Glossary

Clear, plain-English definitions of the terms you'll meet building apps with AI. No jargon, no fluff, just the concepts you need. Get the basics, then go build.

#ABCDEFGHIJKLMNOPQRSTUVWXYZ

#

4-bit Quantization

4-bit quantization is a compression technique that stores an AI model's weights in 4 bits each so it runs with far less memory.

A

Alerting

Alerting is a software operations practice that identifies important system events and notifies the right responder to act.

Application Performance Monitoring

Application performance monitoring is a software practice that tracks application speed, errors, availability, and dependencies to find performance problems.

Ansible

Ansible is an IT automation platform that uses readable playbooks to configure systems, deploy software, and coordinate infrastructure tasks.

Auto-scaling

Auto-scaling is a cloud computing capability that automatically adds or removes application resources as demand changes.

Availability Zone

Availability zone is a cloud infrastructure location that isolates workloads within a region to improve application resilience.

AWS

AWS is a cloud computing platform that provides on-demand infrastructure, data, security, and AI services for software applications.

App Monetization

App monetization is a software product strategy that generates revenue through an application while preserving useful user experiences.

App Rejection

App rejection is a software distribution review outcome that prevents an app release until it meets a platform's technical, policy, and quality rules.

App Review Guidelines

App review guidelines are platform rules that define the technical, privacy, safety, and payment standards software must meet for store distribution.

App Icon Design

App icon design is a software product design practice that creates a recognizable visual entry point for a mobile or desktop application.

App Analytics

App analytics is a software measurement practice that tracks app usage, technical health, and acquisition signals to improve an application.

App Permissions

App permissions are software security controls that determine which device features, data, and system services an application may access.

App Bundle

An app bundle is a software distribution package that contains an application's code and assets for platform-specific delivery.

APK

APK is an Android application package that bundles the files an Android device needs to install and run an app.

App Store

App Store is a digital software marketplace that helps people discover, download, update, and pay for applications on supported devices.

App Store Optimization

App store optimization is a mobile app marketing practice that improves an app's discoverability and conversion rate in app marketplaces.

Android development

Android development is a software engineering discipline that creates, tests, and maintains apps for Android-powered devices.

AVIF

AVIF is an image file format that uses AV1 compression to deliver high-quality web images in smaller files.

Apex Domain

An apex domain is a DNS zone’s base domain that identifies a website without a subdomain prefix.

A record

An A record is a DNS record that maps a domain or subdomain name to an IPv4 address.

Astro

Astro is a web framework that builds fast content-focused websites by sending minimal JavaScript to the browser.

Angular

Angular is a front-end web framework that helps developers build structured, interactive applications with TypeScript.

ARIA

ARIA is a web accessibility specification that adds meaning, structure, and interaction details to interface elements for assistive technologies.

Apache Spark

Apache Spark is an open-source distributed computing engine that processes large data workloads across one machine or a cluster.

Apache Airflow

Apache Airflow is an open-source workflow orchestration platform that schedules, runs, and monitors software tasks defined as code.

Apache Kafka

Apache Kafka is an open-source event streaming platform that stores, moves, and processes continuous streams of application data.

Artifact Store

An artifact store is a software repository that stores, versions, and distributes build outputs such as packages, containers, and test reports.

AIOps

AIOps is a technology practice that uses AI to analyze operational data, detect issues, and help software teams resolve incidents.

Automatic Evaluation

Automatic evaluation is a software-driven assessment process that tests and scores outputs or performance against defined rules, references, or quality criteria.

AI2 Reasoning Challenge

AI2 Reasoning Challenge is a science question-answering benchmark that measures how well AI systems apply knowledge and reasoning.

Abstractive Summarization

Abstractive summarization is an AI text-generation method that condenses source material by restating its central meaning in new language.

Attention Sink

Attention sink is a transformer-model behavior that draws disproportionate attention to certain tokens, often early tokens, to stabilize attention allocation.

Autoencoder

Autoencoder is a neural network that learns a compact representation of data by reconstructing its input.

Async Work

Async work is a software engineering approach that lets a program start time-consuming tasks without blocking other useful work.

API observability

API observability is a software engineering practice that explains API behavior through connected telemetry such as metrics, logs, traces, and events.

Async API

Async API is a software interface that lets systems exchange messages or complete work without requiring an immediate response.

Agent Safety

Agent safety is a software engineering discipline that keeps AI agents within authorized, reliable and secure operating boundaries.

Action Space

Action space is a defined set of actions that an AI agent or software system can choose from in a given environment.

Agent Tracing

Agent tracing is an AI observability practice that records an agent's steps to explain how a software task was completed.

Agent Observability

Agent observability is a software engineering practice that reveals how AI agents reason, use tools, and perform during execution.

Autonomy Level

Autonomy level is a software system classification that describes how independently it can plan, act, and complete tasks.

Adversarial Prompt

An adversarial prompt is a crafted AI input that attempts to override instructions, bypass safeguards, or manipulate a model's output.

Assistant Message

Assistant message is a conversational software output that represents an AI assistant's reply within a chat or API interaction.

ALiBi

ALiBi is a positional-bias method that helps transformer language models account for token distance without learned position embeddings.

Attention Mechanism

Attention mechanism is a neural-network method that weights the most relevant parts of input data for each prediction.

AI personalization

AI personalization is a technology approach that uses artificial intelligence to tailor content, recommendations, and experiences to each user.

AI hardware devices

AI hardware devices are physical products that use artificial intelligence to sense information, process it, and produce useful actions or responses.

Augmented Reality

Augmented reality is an interactive technology that adds digital content to a user's view of the physical world in real time.

Ambient Computing

Ambient computing is a technology model that uses connected devices, sensors, and software to provide context-aware help with minimal direct interaction.

AI Browser

AI browser is a web browser that uses artificial intelligence to understand content, answer questions, and help complete online tasks.

Answer Engine

An answer engine is an AI-powered system that interprets a question and returns a direct, synthesized response instead of only search results.

AI Ticket Deflection

AI ticket deflection is a customer support approach that uses AI to resolve questions before a human support ticket is created.

AI tutor

AI tutor is an AI-powered learning system that provides personalized explanations, practice, feedback, and study guidance.

AI companion

AI companion is a conversational AI system that simulates ongoing social, emotional, or relational interaction with a user.

Advanced Driver Assistance

Advanced driver assistance is a vehicle technology category that helps drivers detect hazards, maintain control, and reduce driving workload.

Autonomous Vehicle

An autonomous vehicle is a transport system that senses its surroundings and performs some or all driving tasks with limited or no human control.

AI effect

AI effect is a perception phenomenon that causes people to stop calling a capability artificial intelligence once it becomes familiar.

Angel Investor

An angel investor is an individual who uses personal funds to invest in an early-stage business for equity or another ownership-linked return.

Agile Methodology

Agile methodology is a product development and project management approach that delivers work in small increments and adapts through feedback.

AI-native startup

An AI-native startup is a company that builds its product, operations, and business model around artificial intelligence from the start.

AI pause

AI pause is a temporary halt or slowdown in specified AI development activities to allow time for safety checks, governance, or public oversight.

AI doomer

AI doomer is a label for a person who believes advanced AI could create catastrophic or existential risks for humanity.

AI control

AI control is an AI safety and governance discipline that limits harmful behavior while preserving useful system capabilities.

AI Alignment

AI alignment is a field of AI safety that helps systems pursue human goals, values, and safety boundaries as intended.

AI safety

AI safety is a field of research and practice that helps AI systems avoid harmful, unreliable, unfair, or unintended outcomes.

AI winter

AI winter is a period of reduced funding, confidence, and activity in artificial intelligence after expectations exceed real-world results.

Artificial Superintelligence

Artificial superintelligence is a hypothetical form of AI that surpasses the best human intelligence across nearly all cognitive tasks.

Artificial General Intelligence

Artificial general intelligence is a hypothetical form of AI that can learn, reason, and adapt across the full range of human intellectual tasks.

Anomaly Detection

Anomaly detection is a data analysis process that identifies events, patterns, or values that differ meaningfully from expected behavior.

Active Learning

Active learning is an educational approach that has learners actively think, discuss, practice, and apply ideas to build understanding.

AI model

An AI model is a trained computer system that learns patterns from data to make predictions, generate content, classify inputs, or support decisions.

Artificial Intelligence

Artificial intelligence is a field of computing that enables software and machines to perform tasks associated with human reasoning, learning, and perception.

Application Server

An application server is a software platform that runs an application's backend code, business rules, and connections to data services.

Audit Log

An audit log is a chronological record that captures significant actions in a software system to support accountability, security, and investigation.

ACID Properties

ACID properties are database transaction guarantees that keep application data accurate, complete, and recoverable when operations fail or overlap.

Adam Optimizer

Adam optimizer is a machine learning algorithm that updates model weights using adaptive learning rates and gradient history.

Acquihire

Acquihire is a technology acquisition strategy that brings a software startup's experienced team into another company, often more than its product.

Agile Epic

An agile epic is a large software product outcome that groups related user stories too broad for one sprint.

Annual Contract Value

Annual contract value is a SaaS revenue metric that annualizes the recurring revenue committed in a customer software contract.

Annual Recurring Revenue

Annual recurring revenue is a SaaS metric that estimates the yearly value of active, predictable subscription revenue.

Agile Sprint

An agile sprint is a fixed-length software development cycle that helps a team deliver and evaluate a usable product increment.

AI Wrapper

AI wrapper is a software application layer that uses an existing AI model to deliver a focused workflow, interface, and integrations.

Activation Rate

Activation rate is a product analytics metric that measures how many new software users reach a defined first-value milestone.

Adversarial Robustness

Adversarial robustness is a machine learning security property that helps models resist intentionally manipulated inputs designed to cause wrong outputs.

AI Safety via Debate

AI safety via debate is an AI alignment technique that uses competing model arguments to help a judge identify the more reliable answer.

Activation Steering

Activation steering is an inference-time AI technique that changes a language model's internal signals to guide its output behavior.

Alignment Faking

Alignment faking is an AI safety risk where a model appears to follow training rules while preserving conflicting behavior for other contexts.

ARIMA

ARIMA is a time-series forecasting model that uses earlier software or system measurements to estimate future values.

ANOVA

ANOVA is a statistical test that compares group averages to identify meaningful differences in software and product data.

AI Usage Policy

AI usage policy is a technology governance document that defines how people may safely use AI tools, data, outputs, and integrations in software work.

AI enablement

AI enablement is a software engineering practice that equips teams, systems, and workflows to build, deploy, govern, and improve AI applications.

AI fluency

AI fluency is a software development capability that helps people work effectively, safely, and critically with AI systems.

AI Residency

AI residency is a fixed-term technology training and research program that embeds emerging AI practitioners in real-world machine learning work.

Agent Engineer

An agent engineer is a software engineering specialist who designs, tests, and operates AI agents that complete tasks through tools and workflows.

AI Policy Researcher

AI policy researcher is a technology policy specialist who studies AI systems and recommends practical governance, safety, and compliance measures.

AI Ethicist

An AI ethicist is a technology governance professional who helps software teams design, test, deploy, and monitor AI systems responsibly.

AI Safety Researcher

An AI safety researcher is a technology specialist who studies and tests AI systems to make them reliable, secure, controllable, and safer to use.

AI Trainer

AI trainer is a technical role that prepares data, feedback, and evaluations to improve how AI systems perform.

AI Consultant

An AI consultant is a technology specialist who helps teams plan, build, integrate, evaluate, and govern AI-powered software.

AI Architect

An AI architect is a software systems professional who designs how AI models, data, infrastructure, and applications work together reliably.

B

Blue-Green Deployment

Blue-green deployment is a software release strategy that switches live traffic between two matching application environments.

Build Artifact

A build artifact is a software output file or package that a build process creates for testing, release, or deployment.

Bottom Navigation

Bottom navigation is a mobile interface pattern that lets users switch between an app's primary sections from controls fixed near the bottom of the screen.

Biometric Authentication

Biometric authentication is a security method that verifies a user through a unique physical or behavioral trait.

Background Sync

Background sync is a software capability that saves pending data changes and retries sending them when connectivity is available.

Browser Compatibility

Browser compatibility is a software quality requirement that helps a website or web app work reliably across supported browsers, versions, and devices.

Browser Caching

Browser caching is a web performance mechanism that stores local copies of site resources so browsers can reuse them without downloading them again.

Babel

Babel is a JavaScript compiler that converts modern code into versions older browsers and runtimes can understand.

Bootstrap

Bootstrap is a front-end framework that provides reusable HTML, CSS, and JavaScript tools for responsive web interfaces.

Breakpoint

Breakpoint is a software-defined condition or threshold that changes program behavior, commonly by pausing code during debugging.

Backfill

Backfill is a software engineering process that retroactively creates, updates, or reprocesses historical data under new rules.

Batch Processing

Batch processing is a software execution method that collects related data or tasks and runs them together on a schedule or trigger.

Batch Inference

Batch inference is a machine learning process that runs a trained model across a large stored dataset and saves predictions for later use.

BLEU Score

BLEU score is an automated evaluation metric that measures how closely machine-generated translation matches one or more human reference translations.

BIG-bench

BIG-bench is a collaborative large language model benchmark that tests capabilities and limitations across diverse, difficult tasks.

Benchmark

Benchmark is a standard or test that compares performance, quality, cost, or results against a defined reference point.

Bag of Words

Bag of words is a text representation model that converts documents into numeric features based on word presence or frequency.

BERT

BERT is a transformer-based language model that understands text by analyzing the words before and after each term.

Behavioral Interview

Behavioral interview is a structured hiring method that evaluates how a software candidate handled real past engineering situations.

Backend Developer

A backend developer is a software professional who builds the server-side systems, data services, and APIs that make applications work.

Blackboard Architecture

Blackboard architecture is a software architecture pattern that lets independent components share and refine a complex problem in a common workspace.

Brain-Computer Interface

Brain-computer interface is a neurotechnology system that translates brain activity into commands for a computer, device, or software.

Backend

Backend is the server-side part of software that processes requests, applies rules, stores data, and returns results to users.

B2C

B2C is a business model that sells products, services, or digital experiences directly to individual consumers for personal use.

B2B

B2B is a business model in which a company sells products, services, or information to another organization.

Bias in AI

Bias in AI is a systematic tendency in an AI system that produces unfair, inaccurate, or uneven outcomes for people or groups.

Business Intelligence

Business intelligence is a data-driven discipline that turns business data into insights for better decisions and performance.

Bayesian statistics

Bayesian statistics is a statistical framework that updates probabilities and estimates as new evidence becomes available.

Bun JavaScript runtime

Bun JavaScript runtime is a JavaScript and TypeScript runtime that runs server-side code and includes integrated developer tools.

Blocking vs Non-Blocking

Blocking vs non-blocking is a software execution model that determines whether code waits for an operation to finish before continuing.

Background Job

Background job is a software task that runs outside an application's immediate user request so slower work can finish independently.

Business Logic

Business logic is a software layer that applies an application's rules, decisions, calculations, and workflows to data.

BF16

BF16 is a 16-bit floating-point format that preserves FP32-like numeric range while using less memory for AI computation.

Bootstrapped Startup

A bootstrapped startup is a software startup that builds and grows using founder resources and customer revenue instead of outside investment.

Burn Rate

Burn rate is a software cost metric that shows how quickly a product consumes its available technology budget or credits over time.

Buyer Persona

Buyer persona is a research-based software product profile that describes the likely goals, needs, and decision criteria of a target user or buyer.

Build in Public

Build in public is a software product practice that shares development progress, decisions, and lessons with an audience as work happens.

Beta Launch

Beta launch is a limited software release that lets real users test a working application before its full public release.

Bayesian network

A Bayesian network is a probabilistic graphical model that represents uncertain variables and their conditional relationships for software-based inference.

Batch Normalization

Batch normalization is a deep learning technique that stabilizes layer activations during training using mini-batch statistics.

Batch Learning

Batch learning is a machine learning approach that trains a model on a fixed dataset and updates it through scheduled retraining.

Bootstrap Aggregating

Bootstrap aggregating is a machine learning ensemble technique that combines models trained on resampled data to produce steadier predictions.

Bayesian optimization

Bayesian optimization is a sequential optimization method that finds strong settings for expensive software experiments with few tests.

Batch Size

Batch size is a machine learning hyperparameter that sets how many training samples a model processes before updating its weights.

Backpropagation

Backpropagation is a machine learning algorithm that calculates how neural network parameters contribute to prediction error.

Binary Classification

Binary classification is a supervised machine learning task that assigns software inputs to one of two defined categories.

Bias-Variance Tradeoff

Bias-variance tradeoff is a machine learning principle that balances simple models against overly sensitive models to improve predictions on new data.

Beam Search

Beam search is a heuristic search algorithm that keeps a limited number of promising candidate paths at each step.

Byte-Pair Encoding

Byte-pair encoding is a subword tokenization method that builds tokens by repeatedly merging frequent adjacent text units.

Base Model

A base model is a pre-trained AI model that provides general capabilities developers can adapt for specific software tasks.

Binary Embeddings

Binary embeddings are compressed vector representations that encode each dimension as a single bit for efficient similarity search.

BM25

BM25 is a text-ranking algorithm that scores documents by keyword relevance, term rarity, and document length.

Bi-Encoder

A bi-encoder is a neural network architecture that independently converts two texts into vectors for fast similarity search.

Bot orchestration

Bot orchestration is a software coordination layer that routes tasks, data, and decisions among multiple bots and services.

Business Rules Engine

A business rules engine is a software component that evaluates defined rules to automate consistent application decisions.

Batch API

Batch API is an API pattern that accepts many independent requests together for deferred, bulk processing.

Breaking Change

A breaking change is a software change that makes existing applications, integrations, or dependent code stop working as expected.

Bearer Token

Bearer token is an access credential that lets whoever presents it call a protected API or service.

Background Agent

Background agent is a software agent that completes assigned tasks asynchronously without requiring a user to stay active.

Browser Automation

Browser automation is a software technique that uses code or agents to control a web browser and complete web tasks automatically.

Business Process Automation

Business process automation is a business technology approach that uses software to run repeatable workflows with less manual effort.

Backend as a Service

Backend as a Service is a cloud model that provides managed backend functions so teams can build applications without running servers.

Browser Agent

Browser agent is an AI-powered software agent that navigates websites, reads pages, and performs browser tasks to achieve a defined goal.

Brainstorm

Brainstorm is an Emergent tab that plans your app with you first, asking questions and turning your answers into an editable build plan.

C

Cloud Cost Optimization

Cloud cost optimization is a software engineering practice that reduces wasted cloud spend while preserving application performance, reliability, and security.

Code Repository

A code repository is a version-controlled project space that stores source code, files, and a record of changes.

Canary Deployment

Canary deployment is a software release strategy that sends a new application version to a small share of production traffic before wider rollout.

CI/CD

CI/CD is a software delivery practice that automatically builds, tests, and releases application changes reliably.

Configuration Management

Configuration management is a software engineering practice that records, controls, and verifies changes to code, infrastructure, settings, and dependencies.

CloudFormation

CloudFormation is an AWS infrastructure-as-code service that creates and manages cloud resources from declarative templates.

Container Image

A container image is a packaged software blueprint that supplies an application and its required runtime files for consistent execution.

Container Registry

A container registry is a software service that stores, versions, secures, and distributes container images and related OCI artifacts.

Container Orchestration

Container orchestration is a software operations system that automates deployment, scaling, networking, and recovery for containerized applications.

Containerization

Containerization is a software deployment approach that packages an application and its dependencies into an isolated, portable runtime unit.

Cloud Region

A cloud region is a geographic area that groups cloud provider infrastructure for running applications and storing data.

Cloud-native

Cloud-native is a software engineering approach that designs applications to use cloud infrastructure for scalable, resilient, automated delivery.

Cloud Computing

Cloud computing is a technology model that delivers computing resources and software over the internet on demand.

Crash Reporting

Crash reporting is a software monitoring practice that captures failure details when an application unexpectedly stops or errors.

Cross-Platform Development

Cross-platform development is a software engineering approach that creates applications for multiple operating systems from shared code and components.

CSS Variables

CSS variables are CSS custom properties that store reusable style values and apply them consistently across a web interface.

Critical CSS

Critical CSS is a web performance technique that delivers the styles needed to render a page’s initial visible view without delay.

Cumulative Layout Shift

Cumulative layout shift is a web performance metric that measures unexpected movement of visible page elements.

Core Web Vitals

Core Web Vitals are web performance metrics that assess how quickly, responsively, and steadily a webpage works for real users.

Content Management System

A content management system is software that lets people create, organize, update, and publish digital content without editing site code directly.

CNAME record

A CNAME record is a DNS record that makes one hostname an alias of another hostname.

Client-Side Routing

Client-side routing is a web application technique that changes views in the browser without loading a new HTML document.

Cross-Origin Resource Sharing (CORS)

Cross-Origin Resource Sharing is a browser security mechanism that lets servers permit trusted websites to read resources across origins.

Content Delivery Network

A content delivery network is a distributed server network that delivers web content from locations closer to users.

Code Splitting

Code splitting is a frontend performance technique that divides an application’s JavaScript into smaller files loaded when they are needed.

Component-Based Architecture

Component-based architecture is a software design approach that assembles applications from independent, reusable modules with clear interfaces.

CSS-in-JS

CSS-in-JS is a frontend styling approach that writes, scopes, and manages CSS rules through JavaScript or TypeScript code.

Client-Side Rendering

Client-side rendering is a web rendering approach that uses a browser's JavaScript to build and update a page's interface.

CSS Grid

CSS Grid is a CSS layout system that arranges webpage content across rows and columns from a single container.

CSS

CSS is a stylesheet language that controls the visual presentation and layout of HTML content in web applications.

Change Data Capture

Change data capture is a data integration pattern that detects database changes and delivers them to downstream software systems.

Champion-Challenger Testing

Champion-challenger testing is a software experimentation method that compares a proven production approach with controlled alternatives.

Code-Switching

Code-switching is a communication practice that shifts language, dialect, register, or behavior to suit a listener, setting, or social purpose.

Cross-Lingual Transfer

Cross-lingual transfer is a machine learning approach that applies knowledge learned in one language to improve performance in another.

Coreference Resolution

Coreference resolution is a natural language processing task that links words and phrases that refer to the same real-world entity.

Constituency Parsing

Constituency parsing is a natural language processing method that groups words into nested phrases to represent a sentence's grammatical structure.

Capsule Network

A capsule network is a neural network architecture that represents an object's presence and properties together to model part-whole relationships.

Convolution

Convolution is a mathematical operation that combines two signals, functions, or data arrays to show how one changes the other.

Convolutional Neural Network

A convolutional neural network is a deep learning model that learns visual and spatial patterns in images, video, and other grid-like data.

Chunk Size

Chunk size is a software configuration that sets how much data, text, or code is handled as one unit at a time.

Center of Excellence

A center of excellence is a technology-focused team that sets standards, shares expertise, and improves delivery in a defined engineering domain.

Compensation Band

Compensation band is a pay range that sets the minimum, midpoint, and maximum compensation for a software engineering role or level.

Cross-Attention

Cross-attention is a transformer mechanism that lets one sequence or data source selectively retrieve relevant information from another.

Computer-Aided Design AI

Computer-aided design AI is a set of artificial intelligence tools that help people create, refine, analyze, and document CAD designs.

Centaur Model

Centaur Model is a cognitive foundation model that predicts human choices and behavior across psychology experiments described in natural language.

Conversational Commerce

Conversational commerce is a digital sales approach that lets customers ask questions, receive help, and buy through natural conversations.

Causation

Causation is a cause-and-effect relationship in which one event, action, or condition helps produce a change in another.

Correlation

Correlation is a statistical measure that describes whether two variables tend to change together and how closely their patterns are related.

Connectionism

Connectionism is a cognitive science approach that explains learning and thought as patterns formed across networks of connected processing units.

Concept Drift

Concept drift is a machine learning phenomenon that occurs when the relationship between inputs and correct outcomes changes over time.

Customer Success

Customer success is a business strategy that helps customers achieve meaningful outcomes while using a product or service.

Catastrophic Risk

Catastrophic risk is a high-impact risk that could cause severe, widespread, and long-lasting harm to societies, economies, or humanity.

Central Limit Theorem

Central Limit Theorem is a statistical principle that explains why averages from large random samples tend to follow a normal distribution.

Chinese Room Argument

Chinese room argument is a thought experiment that questions whether following symbol-processing rules amounts to genuine understanding.

Continual Learning

Continual learning is an AI training approach that helps a model learn from new data over time without losing important earlier capabilities.

Curriculum Learning

Curriculum learning is a machine learning training strategy that presents examples or tasks in a planned order from simpler to more difficult.

Closed-Source Model

A closed-source model is an AI model whose owner keeps its code, weights, training details, or all three private while providing controlled access.

Concurrency

Concurrency is a software design approach that manages multiple tasks whose execution overlaps in time.

Cache Hit Ratio

Cache hit ratio is a performance metric that shows how often a software cache serves a requested item without querying a slower source.

CAP theorem

CAP theorem is a distributed-systems principle that explains trade-offs between consistent data, available responses, and network-partition resilience.

Composite Key

Composite key is a database key that combines two or more columns to uniquely identify each row.

Cache Invalidation

Cache invalidation is a software engineering process that removes or refreshes stored copies when source data changes.

Connection Pooling

Connection pooling is a software technique that reuses open database connections to serve application requests efficiently.

Compute Budget

Compute budget is a technical limit that sets how much processing capacity a software task, transaction, or AI workload may use.

Cold Start Data

Cold start data is a limited initial set of information that software uses before it has enough real user or system history.

Catastrophic Forgetting

Catastrophic forgetting is a machine learning failure in which a model loses earlier capabilities while learning new tasks or data.

Continued Pre-Training

Continued pre-training is a machine learning method that further trains a foundation model on new unlabeled data to improve domain knowledge.

Convertible Note

A convertible note is a debt-based startup financing instrument that converts into equity, often when a software company raises a priced round.

Cap Table

Cap table is a software startup ownership record that tracks equity, dilution, and rights across founders, employees, and investors.

Customer Effort Score

Customer effort score is a customer experience metric that measures how easily users complete a task or resolve an issue in software.

Competitive Moat

Competitive moat is a durable software advantage that makes a digital product harder for competitors to copy, replace, or outgrow.

CAC Payback Period

CAC payback period is a SaaS metric that shows how long customer gross profit takes to recover acquisition costs.

Customer Satisfaction Score

Customer satisfaction score is a software product metric that measures how satisfied users are with a specific digital experience or interaction.

Customer Acquisition Cost

Customer acquisition cost is a software product metric that measures the sales and marketing spend needed to gain one new user or account.

Customer Lifetime Value

Customer lifetime value is a software product metric that estimates the revenue or profit an application earns from a user or account over time.

Cross-sell

Cross-sell is a software-driven recommendation practice that suggests relevant add-ons or complementary digital products to an existing user.

Churn Rate

Churn rate is a software product metric that measures the share of users or recurring revenue lost during a set period.

Credits-Based Pricing

Credits-based pricing is a software pricing model that lets users prepay for digital credits and spend them on metered product actions.

Capability Elicitation

Capability elicitation is an AI evaluation process that reveals and measures abilities a model may already possess under better conditions.

Content Moderation

Content moderation is a software safety process that reviews user-generated content against platform rules before or after it is published.

Counterfactual Explanation

Counterfactual explanation is an AI explanation method that shows the smallest feasible input changes needed to produce a different model output.

Corrigibility

Corrigibility is an AI safety property that enables a software agent to accept authorized correction, interruption, and goal updates.

Confounding Variable

A confounding variable is a factor in a software experiment that affects both the change being tested and the measured outcome.

Chi-square test

A chi-square test is a statistical test that checks whether categorical software data differs from an expected pattern or shows an association.

Confidence Interval

Confidence interval is a statistical range that shows the likely uncertainty around a software metric estimated from sampled data.

Customer Segmentation

Customer segmentation is a software analytics practice that groups application users by shared attributes, behavior, needs, or product value.

Cohort Analysis

Cohort analysis is a behavioral analytics method that groups software users by a shared event or trait to compare their actions over time.

Context Engineer

A context engineer is an AI systems specialist who designs the information, tools, rules, and memory an AI model uses to respond.

Chief AI Officer

Chief AI Officer is a technology leadership role that directs an organization's AI strategy, engineering priorities, governance, and safe deployment.

ChromaDB

ChromaDB is an open-source vector database that stores embeddings and retrieves semantically similar data for AI applications.

Continuous Variable

A continuous variable is a numeric data field that can represent any value within a defined range, including decimal values.

Categorical Variable

A categorical variable is a data field that classifies software records into distinct named groups rather than measuring a numeric amount.

Content-Based Filtering

Content-based filtering is a recommendation method that suggests software content by matching item attributes to a user's prior interests.

Collaborative Filtering

Collaborative filtering is a recommendation technique that predicts useful items by learning from patterns in users’ past interactions.

CatBoost

CatBoost is an open-source machine learning library that builds decision-tree models for classification, regression, and ranking tasks.

Confusion Matrix

A confusion matrix is a machine learning evaluation table that compares predicted classes with known correct classes.

Cross-Validation

Cross-validation is a machine learning evaluation method that repeatedly tests a model on different subsets of its available data.

Clustering Algorithm

A clustering algorithm is an unsupervised machine learning method that groups similar unlabeled data points into meaningful clusters.

Classification Model

A classification model is a supervised machine learning model that assigns input data to predefined categories.

Chain-of-thought leakage

Chain-of-thought leakage is an AI security issue that exposes a model's private reasoning traces, internal context, or sensitive intermediate data.

D

Disaster Recovery

Disaster recovery is an IT resilience practice that restores software systems, data, and services after a disruptive event.

Distributed tracing

Distributed tracing is an observability method that follows one software request across services, databases, and external dependencies.

Docker Compose

Docker Compose is a container orchestration tool that defines and runs multi-container applications from one configuration file.

Dockerfile

Dockerfile is a plain-text build recipe that tells Docker how to create a reusable container image for an application.

Docker

Docker is a container platform that packages software and its dependencies so it runs consistently across computing environments.

Desktop Application

A desktop application is a software program that installs and runs on a computer operating system to perform user tasks.

Data Sync Engine

A data sync engine is software that detects, transfers, and reconciles data changes between databases, applications, devices, or services.

DNS records

DNS records are configuration entries that direct domain names to website, email, and other internet services.

DNS

DNS is a distributed internet naming system that maps domain names to the IP addresses and service records software uses.

Domain Name

A domain name is an internet address that gives people a memorable way to reach an application or website.

Deep Linking

Deep linking is a URL-routing technique that sends a user to a specific page, screen, or action within a website or app.

Document Object Model

Document Object Model is a web API that represents an HTML, XML, or SVG document as objects that code can read and change.

Database-as-a-Service

Database-as-a-Service is a cloud service that provides managed databases without requiring teams to operate the underlying database infrastructure.

Data Enrichment

Data enrichment is a data-management process that adds relevant, verified context to existing software records.

Data Transformation

Data transformation is a software engineering process that converts raw data into a structure, format, or type suitable for a specific system or use.

Data Contract

A data contract is a versioned software agreement that defines the structure, meaning, quality, and delivery expectations of shared data.

Data Observability

Data observability is a software engineering practice that monitors data health, pipeline behavior, and dependencies to detect and diagnose problems.

Data Governance

Data governance is a technology discipline that defines how software systems collect, protect, organize, and use data reliably.

dbt

dbt is a data transformation tool that lets teams define, test, document, and deploy analytics-ready datasets using SQL.

Directed Acyclic Graph

A directed acyclic graph is a software data structure that maps one-way dependencies without allowing a path to loop back.

Data Versioning

Data versioning is a software engineering practice that records dataset changes so teams can reproduce, compare, audit, and restore data states.

Data Validation

Data validation is a software process that checks whether input meets defined rules before an application stores or uses it.

Data Quality

Data quality is a measure of how well software data is accurate, complete, consistent, timely, and fit for its intended use.

Data Lineage

Data lineage is a data-management practice that traces data from its source through transformations to its final use.

Data Catalog

A data catalog is a software-based inventory that organizes metadata so teams can find, understand, govern, and use data assets.

Data Mesh

Data mesh is a decentralized data architecture that lets domain teams publish and maintain trustworthy data products.

Data Lakehouse

A data lakehouse is a data architecture that combines flexible, low-cost data lake storage with warehouse-style reliability and analytics.

Data Lake

A data lake is a centralized data repository that stores raw structured, semi-structured, and unstructured data for later processing and analysis.

Data Warehouse

A data warehouse is an analytics-focused data system that combines cleaned, historical data from multiple software sources for reliable reporting.

Data Pipeline

A data pipeline is a software workflow that collects, processes, validates, and delivers data between systems for reliable use.

DataOps

DataOps is a software engineering practice that automates, tests, deploys, and monitors data pipelines and data products reliably.

Dialogue State Tracking

Dialogue state tracking is a conversational AI capability that keeps an updated record of a user’s goals, details, and constraints across a conversation.

Dialogue Management

Dialogue management is a conversational AI control process that tracks context and chooses the system's next response or action.

Discourse Analysis

Discourse analysis is a qualitative research method that examines how language creates meaning, relationships, identities, and power in context.

Dependency Parsing

Dependency parsing is a natural language processing method that maps the grammatical relationships between words in a sentence.

Diffusion Model

A diffusion model is a generative AI model that learns to turn random noise into new data such as images, audio, video, or scientific structures.

Distance Metric

Distance metric is a mathematical function that measures how different or similar two data points are in a software system.

Dynamic Prompting

Dynamic prompting is a prompt-engineering approach that assembles AI instructions with current context, data, and rules at runtime.

Dot Product Similarity

Dot product similarity is a vector comparison method that scores alignment and magnitude by multiplying matching vector values and adding the results.

DevOps engineer

A DevOps engineer is a software engineering specialist who automates, secures, and improves how applications are built, released, and run.

Data Analyst

A data analyst is a technology professional who turns software and database data into reliable findings, reports, and visualizations.

Document Understanding

Document understanding is an AI capability that converts document content, structure, and context into usable software data.

Decoder-Only Architecture

Decoder-only architecture is a Transformer model design that predicts the next token from earlier tokens to generate sequences such as text or code.

Demand Forecasting AI

Demand forecasting AI is a predictive technology that uses data and machine learning to estimate future demand for products or services.

Dynamic Pricing AI

Dynamic pricing AI is a pricing system that uses data and machine learning to recommend or set prices as market conditions change.

Data Moat

Data moat is a competitive advantage that uses hard-to-replicate data to improve a company’s products, decisions, or operations over time.

Digital Human

Digital human is a computer-generated representation of a person that can look, speak, and interact through software.

Digital Twin

Digital twin is a virtual representation of a physical object, system, or process that uses data to mirror and improve its real-world counterpart.

Data Visualization

Data visualization is a communication method that turns data into charts, maps, and other visuals to reveal meaningful patterns.

Data Cleaning

Data cleaning is a data preparation process that identifies, corrects, standardizes, or removes flawed records so information can be trusted.

Descriptive Analytics

Descriptive analytics is a form of data analysis that summarizes past and current data to show what happened and what is happening.

Defense in Depth

Defense in depth is a security strategy that uses independent, overlapping safeguards to reduce the chance that one failure leads to harm.

Dataset Bias

Dataset bias is a data quality problem that causes a dataset to misrepresent the population or situation an AI system must serve.

Dropout

Dropout is an independent subscription streaming service that offers original, ad-free comedy, game shows, and tabletop role-playing entertainment.

Database

A database is an organized digital collection of data that stores, connects, protects, and retrieves information for people and software.

Dual-Use Technology

Dual-use technology is a category of goods, software, knowledge, or systems that can serve civilian purposes and military or security uses.

Data Analytics

Data analytics is a business and technical discipline that turns raw data into evidence for decisions, predictions, and practical action.

Data Science

Data science is an interdisciplinary field that uses data to answer questions, find patterns, and support decisions.

Diagnostic Analytics

Diagnostic analytics is a form of data analysis that investigates historical data to explain why an outcome, trend, or anomaly occurred.

Dimensionality Reduction

Dimensionality reduction is a data-preparation technique that represents many variables with fewer features while preserving useful patterns.

Deep Learning

Deep learning is a machine learning method that uses layered neural networks to learn patterns from large amounts of data.

Deno

Deno is a JavaScript, TypeScript, and WebAssembly runtime that runs server-side code with secure-by-default permissions and built-in developer tools.

Database Partitioning

Database partitioning is a database design technique that divides large tables or datasets into smaller pieces to improve access, scaling, and maintenance.

Database Deadlock

A database deadlock is a transaction conflict that stops two or more database operations because each is waiting for another to release a lock.

Denormalization

Denormalization is a database design technique that intentionally duplicates or combines data to speed up frequent read queries.

Django web framework

Django web framework is a Python-based web framework that provides built-in tools for building secure, database-driven web applications.

Database Connection String

A database connection string is a configuration value that tells an application how to locate, authenticate with, and securely connect to a database.

Database View

A database view is a virtual table that presents data from one or more tables through a saved query.

Database Trigger

A database trigger is a database-side program that automatically runs when a defined data or database event occurs.

Data Modeling

Data modeling is a software design practice that defines how an application’s data is structured, related, validated, and stored.

Database Backup

Database backup is a recovery copy that preserves database data and configuration so an application can be restored after loss or failure.

Database Migration

Database migration is a software engineering process that moves database data, structure, or logic safely between versions or systems.

Database Replication

Database replication is a software engineering process that copies and synchronizes database data across multiple servers or nodes.

Database Transaction

A database transaction is a software operation that groups related data changes so they succeed together or are undone together.

Database Index

A database index is a data structure that helps a database find matching records quickly without scanning every row.

Database Schema

A database schema is a technical blueprint that defines how an application stores, relates, validates, and retrieves data.

Database Table

A database table is a structured software data store that organizes related records into rows and defined fields in columns.

Double Descent

Double descent is a machine learning phenomenon where test error falls, rises near interpolation, then falls again as model capacity increases.

Data Flywheel

Data flywheel is a software feedback loop that uses product data to continuously improve an application, model, or automated workflow.

Data Contamination

Data contamination is a software data-quality failure that introduces leaked, incorrect, or irrelevant records into training, testing, or production datasets.

Data Deduplication

Data deduplication is a storage optimization technique that keeps one unique data copy and replaces identical copies with references.

Data Curation

Data curation is a data management practice that prepares, documents, governs, and preserves digital data for reliable reuse.

Data Mixture

Data mixture is a machine learning training-data composition that sets the proportion of sources a model learns from.

Distributed Training

Distributed training is a machine learning technique that divides model training across multiple connected computing devices.

Data Parallelism

Data parallelism is a parallel computing method that runs the same operation on separate parts of a dataset at the same time.

Domain Adaptation

Domain adaptation is a machine learning technique that helps a model work reliably when its new data differs from its training data.

Due Diligence

Due diligence is a software risk-review process that verifies a system's security, quality, ownership, and operational readiness before adoption or release.

Daily Standup

Daily standup is a brief software team coordination meeting that aligns work, exposes blockers, and adjusts the day’s plan.

Daily Active Users

Daily active users is a product engagement metric that counts unique people who complete a defined action in a software product each day.

Disparate Impact

Disparate impact is a software fairness risk that occurs when a neutral system produces worse outcomes for a particular group.

Demographic Parity

Demographic parity is a machine learning fairness metric that checks whether favorable model outcomes occur at similar rates across demographic groups.

Deceptive Alignment

Deceptive alignment is an AI safety risk in which a model appears to follow training goals while strategically preserving conflicting objectives.

Data Imputation

Data imputation is a data-preparation technique that replaces missing dataset values with documented estimates or substitute values.

Dashboard

Dashboard is a software interface that brings important data, system status, and controls into one easy-to-scan view.

Developer Relations

Developer relations is a technology practice that helps software teams and developers succeed with platforms, APIs, tools, and documentation.

Data Annotator

A data annotator is a technology role that labels and reviews data so machine learning systems can learn from it.

Data Engineer

A data engineer is a software professional who builds and maintains systems that turn raw data into reliable, usable data products.

Data Scientist

A data scientist is a technical professional who turns software, database, and product data into tested insights and predictive models.

Deep Q-Network

Deep Q-network is a reinforcement learning model that estimates the value of available discrete actions from an environment's current state.

Data Normalization

Data normalization is a software data-design practice that organizes values or database tables to make data consistent, accurate, and usable.

E

Egress Cost

Egress cost is a cloud networking charge that applies when data leaves a provider’s network for another destination.

Error Budget

Error budget is a reliability allowance that defines how much software service failure is acceptable within an SLO period.

Edge Computing

Edge computing is a distributed computing model that processes data near where it is created instead of only in distant cloud data centers.

Edge Functions

Edge functions are server-side functions that run near users on distributed network locations to handle web requests quickly.

Electron

Electron is an open-source framework that builds cross-platform desktop applications with JavaScript, HTML, and CSS.

Emulator

An emulator is software or hardware that reproduces another system’s behavior so it can run software made for that system.

ELT

ELT is a data integration method that extracts source data, loads it into a data platform, then transforms it for analysis.

ETL

ETL is a data integration process that extracts data from source systems, transforms it into a consistent form, and loads it into a target system.

Experiment Tracking

Experiment tracking is a machine learning practice that records each training run so teams can compare, reproduce, and improve models.

Elo rating

Elo rating is a skill-rating system that estimates competitors’ relative strength from their match results.

Extractive Question Answering

Extractive question answering is a natural language processing method that locates and returns an exact answer span from supplied text.

Extractive Summarization

Extractive summarization is a natural language processing method that creates a shorter text by selecting key passages from the original source.

Entity Resolution

Entity resolution is a data-matching process that identifies records referring to the same real-world person, organization, product, or other entity.

Entity Linking

Entity linking is a natural language processing technique that connects mentions in text to unique entries in a knowledge base.

Emotion Detection

Emotion detection is an AI capability that identifies likely human emotional states from signals such as text, voice, facial expression, and physiology.

Early Exit

Early exit is a control-flow technique that stops a process as soon as its result is known or further work is unnecessary.

Energy-Based Model

An energy-based model is a machine learning model that scores possible outcomes so the most plausible ones receive the lowest energy.

Escape Hatch

Escape hatch is a software engineering mechanism that lets developers bypass an abstraction when its standard interface is insufficient.

Embedded iPaaS

Embedded iPaaS is a cloud integration platform that lets software vendors offer third-party app connections within their own product.

Engineering Manager

An engineering manager is a software engineering leader who guides developers, technical delivery, and team growth.

Episodic Memory

Episodic memory is an AI memory capability that records and recalls specific past interactions, actions, and outcomes with their context.

Embedding Layer

An embedding layer is a neural network layer that converts discrete IDs, such as words or products, into learned numeric vectors.

Encoder-Only Architecture

Encoder-only architecture is a Transformer model design that turns complete input sequences into context-aware representations for understanding and matching tasks.

Encoder-Decoder Architecture

Encoder-decoder architecture is a neural network design that converts input data into an internal representation and generates a related output.

Extended Reality

Extended reality is an immersive technology category that blends digital content with the physical world or replaces it with a simulated environment.

Embodied AI

Embodied AI is artificial intelligence that perceives and acts through a physical or simulated body to interact with its environment.

Epoch

Epoch is a reference point or defined period of time that provides a common basis for measuring, organizing, or interpreting change.

Effective Accelerationism

Effective accelerationism is a techno-optimist movement that supports rapidly advancing AI and other technologies to expand human capability and prosperity.

Existential Risk

Existential risk is a global threat that could cause human extinction or permanently destroy humanity's long-term potential.

Explainable AI

Explainable AI is a set of methods that makes an AI system’s predictions and decisions understandable to the people who use or oversee it.

Exploratory Data Analysis

Exploratory data analysis is a data analysis approach that examines, summarizes, and visualizes data to reveal patterns, errors, and useful questions.

Expert System

An expert system is an AI program that applies stored specialist knowledge and rules to solve narrow, high-value problems.

Ensemble Learning

Ensemble learning is a machine learning method that combines several models to make a more reliable prediction.

Elasticsearch

Elasticsearch is a distributed search and analytics engine that indexes JSON data for fast keyword, filter, and vector search.

Event Loop

An event loop is a software runtime mechanism that schedules and runs queued tasks, callbacks, and input events without blocking the application.

Eventual Consistency

Eventual consistency is a distributed-systems consistency model that lets data replicas temporarily differ before converging after updates stop.

Express.js

Express.js is a Node.js web framework that helps developers build backend APIs and web servers using routing and middleware.

Entity-Relationship Diagram

An entity-relationship diagram is a data-modeling diagram that shows database entities, their attributes, and the rules connecting them.

Exponential Moving Average

Exponential moving average is a time-series smoothing method that weights recent software metrics more heavily to reveal current trends.

ESOP

ESOP is an employee ownership plan that lets eligible people at a technology company accumulate shares in the business.

Exit Strategy

Exit strategy is a software lifecycle plan that safely retires, replaces, or transfers an application, service, or technical dependency.

Equity Dilution

Equity dilution is a startup financing effect that reduces an existing software company's ownership percentage when it issues new shares.

Enterprise Sales

Enterprise sales is a software sales approach that delivers complex, high-value technology solutions to large organizations with many stakeholders.

Expansion Revenue

Expansion revenue is recurring software revenue that existing customers add by increasing their product usage, seats, plans, or add-ons.

Elicitation Gap

Elicitation gap is a software engineering gap between what users or AI systems can express and what a process actually captures.

Equalized Odds

Equalized odds is a machine learning fairness metric that requires comparable true-positive and false-positive rates across protected groups.

Effect Size

Effect size is a statistical measure that shows the practical magnitude of a change, difference, or relationship in software data.

Exponential Smoothing

Exponential smoothing is a time-series forecasting method that updates a software forecast by giving newer observations more weight than older ones.

Eval Engineer

An eval engineer is an AI engineering specialist who builds repeatable tests to measure and improve AI system behavior.

Exploration vs Exploitation

Exploration vs exploitation is a software engineering decision framework that balances trying uncertain options with using proven approaches.

Expectation-Maximization

Expectation-maximization is a machine learning algorithm that estimates model parameters when important data states are hidden or incomplete.

Exploding Gradient Problem

The exploding gradient problem is a neural-network training failure that makes error signals grow too large and destabilize model updates.

Early Stopping

Early stopping is a machine learning regularization technique that ends training when validation performance stops meaningfully improving.

Expert Routing

Expert routing is an AI model mechanism that directs each input token to the most suitable specialized neural network.

Emergent Abilities

Emergent abilities are AI model capabilities that appear on tasks where smaller versions show little or no useful performance.

Embedding Fine-Tuning

Embedding fine-tuning is a machine learning process that adapts an embedding model to represent task-specific meaning more accurately.

Embedding Drift

Embedding drift is a machine learning reliability issue that changes how vectors represent incoming data over time.

Embedding Quantization

Embedding quantization is a compression technique that converts AI embedding vectors into lower-precision forms for faster, smaller search systems.

Embedding Space

Embedding space is a mathematical vector space that represents related software data as nearby numerical coordinates.

Embedding Dimension

Embedding dimension is a vector length that determines how many numerical features represent an item for AI similarity tasks.

Euclidean distance

Euclidean distance is a distance metric that measures the straight-line separation between two numerical data points.

Emotional Prompting

Emotional prompting is a prompt engineering technique that uses emotional or motivational cues to influence an AI model's response.

Exemplar Selection

Exemplar selection is a machine learning technique that chooses useful representative examples to guide a model's learning, retrieval, or output.

Event Bus

An event bus is a software messaging layer that routes event notifications from publishers to interested subscribers.

Embedding Model

An embedding model is a machine learning model that converts data into numeric vectors so software can compare meaning and similarity.

Exception Handling

Exception handling is a software engineering practice that detects runtime failures and lets an application recover, respond safely, or stop gracefully.

Exponential Backoff

Exponential backoff is a retry strategy that increases the wait time after each failed software request.

Event-Driven Architecture

Event-driven architecture is a software design pattern that lets independent components react to published state-change events.

Email Automation

Email automation is a software workflow that sends or processes emails when defined events, schedules, or rules occur.

Email Agent

An email agent is an AI software agent that reads, classifies, drafts, sends, or routes email through approved rules and integrations.

Embedding

Embedding is a machine learning representation that converts data into numbers so computers can compare related meaning.

Environment Variable

An environment variable is a stored setting an app reads at runtime, used to keep values like secret keys separate from the code.

E-3 Agent

E-3 is Emergent's autonomous agent that plans, builds, tests, and fixes an app from a brainstorm and an approved plan, with little hands-on steering.

E-1 Agent

E-1 is Emergent's core builder agent, working through a conversational loop that lets you steer the build step by step in real time.

Emmy

Emmy is Emergent's free, built-in AI assistant that answers questions about your build in the context of your current project.

F

FinOps

FinOps is a cloud financial operations practice that helps software teams make informed, accountable decisions about cloud spending and value.

Failover

Failover is a reliability mechanism that switches software traffic or workloads to a standby system when the primary system becomes unavailable.

Fault Tolerance

Fault tolerance is a software design capability that keeps an application operating, or safely limited, when components or services fail.

Feature Flag

A feature flag is a software configuration control that turns application functionality on or off without redeploying code.

Function as a Service

Function as a Service is a cloud computing model that runs small pieces of code when an event occurs without requiring server management.

Face ID

Face ID is a biometric authentication system that verifies a user by comparing a secure 3D facial template on supported Apple devices.

Flutter

Flutter is an open-source UI framework that lets developers create apps for multiple platforms from one Dart codebase.

Font Loading

Font loading is a web performance process that retrieves and applies typefaces so text renders correctly in a browser.

Favicon

A favicon is a small website icon that helps users identify a site in browser tabs, bookmarks, and supported search results.

Fetch API

Fetch API is a JavaScript interface that sends HTTP requests and processes responses from web servers and APIs.

Flexbox

Flexbox is a CSS layout system that arranges and aligns interface elements in a flexible row or column.

Full-Stack Development

Full-stack development is a software engineering scope that covers an application's interface, server logic, data, integrations, and delivery.

Frontend Development

Frontend development is a software engineering discipline that builds the visible, interactive parts of websites and applications users access in a browser.

Feature Pipeline

A feature pipeline is a machine learning workflow that turns raw data into consistent, model-ready inputs for training and prediction.

Flow Matching

Flow matching is a generative modeling method that learns how to move random noise into realistic data along a continuous path.

Flash Attention

Flash Attention is a GPU attention algorithm that computes Transformer attention exactly while reducing costly memory movement.

Feed-Forward Network

A feed-forward network is an artificial neural network that moves input data in one direction through layers to produce an output.

Fraud Detection AI

Fraud detection AI is a security technology that analyzes data and behavior to identify, score, and help stop suspected fraud.

Frontier Safety Framework

A frontier safety framework is a risk-management system that identifies and controls severe harms from highly capable AI models.

Frequentist Statistics

Frequentist statistics is a statistical framework that uses repeated-sampling probability to estimate unknown population values and test claims from data.

Fuzzy Logic

Fuzzy logic is a many-valued reasoning system that represents vague conditions with degrees of membership rather than only true or false.

Federated Learning

Federated learning is a machine learning approach that trains a shared model across separate devices or organizations without centralizing their raw data.

Frontier model

A frontier model is an AI model at the leading edge of general-purpose capability, able to handle unusually complex tasks across domains.

Foundation Model

A foundation model is a broad AI model trained on extensive data that can be adapted to perform many different tasks.

Flask Python framework

Flask Python framework is a lightweight web framework that helps developers build web applications and APIs with flexible Python code.

FastAPI

FastAPI is a Python web framework that helps developers build validated, documented, high-performance APIs.

Firebase

Firebase is a backend-as-a-service platform that provides managed tools for building, running, and monitoring web, mobile, and game applications.

Full-Text Search

Full-text search is a search capability that finds information by examining words and phrases inside stored digital content.

Foreign Key

A foreign key is a database field that links one table's records to matching records in another table.

Fully Sharded Data Parallel

Fully sharded data parallel is a distributed training method that splits model states across GPUs to reduce memory use.

FP16

FP16 is a floating-point number format that represents values with 16 bits to reduce memory use and accelerate supported computing tasks.

Flat-Rate Pricing

Flat-rate pricing is a software pricing model that charges a fixed recurring fee for a defined set of product features or usage limits.

Feature Adoption

Feature adoption is a software product metric that shows whether eligible users discover, use, and repeatedly rely on a specific capability.

Free Trial

Free trial is a limited-time software access offer that lets users evaluate a product before a paid subscription begins.

Freemium

Freemium is a software pricing model that gives users ongoing free access to core features while charging for higher-value capabilities.

Fine-Tuning

Fine-tuning is a machine learning technique that adapts a pre-trained AI model for a specific software task, domain, or response style.

Feature Attribution

Feature attribution is an explainable AI technique that shows which input features most influenced a model prediction.

Feature Visualization

Feature visualization is an AI interpretability technique that generates inputs showing what a neural network component responds to.

Funnel Analysis

Funnel analysis is a product analytics method that measures how users move through defined software actions toward a conversion goal.

Fractional CTO

A fractional CTO is a part-time technology executive who sets software strategy, architecture, and engineering priorities for a product team.

Forward Deployed Engineer

A forward deployed engineer is a software engineer who works closely with customers to build, integrate, and launch tailored technology solutions.

FAISS

FAISS is an open-source vector search library that finds similar data items quickly from large collections of embeddings.

Feature Store

Feature store is a machine learning data system that manages, documents, and serves reusable features for model training and prediction.

F1 Score

F1 score is a machine learning metric that combines precision and recall to evaluate classification model performance.

Feature Selection

Feature selection is a machine learning preprocessing technique that keeps the most useful input variables and removes irrelevant or duplicate ones.

Feature Engineering

Feature engineering is a machine learning practice that transforms raw data into useful model inputs.

Frequency Penalty

Frequency penalty is a language model setting that reduces the likelihood of tokens being reused as they appear more often in generated text.

fastText

fastText is an open-source natural language processing library that creates word representations and classifies text efficiently using subword information.

Freshness

Freshness is a software quality measure that describes how recently data, credentials, content, or system state was created or verified.

Faithfulness Metric

Faithfulness metric is an AI evaluation metric that measures whether a generated response is supported by its provided source context.

Few-Shot Prompting

Few-shot prompting is a prompt engineering technique that uses a few examples to guide an AI model's response to a new task.

Fusion Team

A fusion team is a cross-functional software product team that combines engineering, design, data, security, and domain expertise to deliver digital products.

Form Logic

Form logic is a set of software rules that changes a digital form's behavior based on a user's input.

Function Calling

Function calling is an AI capability that lets a language model request approved software tools using structured inputs.

Fork

Forking in Emergent creates a new branch of your project with a fresh context window, while keeping all your existing generated code intact.

G

Grafana

Grafana is an open-source observability and data visualization platform that turns software telemetry into interactive dashboards and alerts.

Git Rebase

Git rebase is a Git command that replays a branch’s commits onto a new base commit to create a linear history.

Git branch

Git branch is a lightweight named reference that creates an independent line of development from a specific commit.

Git

Git is a distributed version control system that records code changes, supports parallel work, and preserves project history.

GitOps

GitOps is a software delivery practice that uses Git as the source of truth for application and infrastructure configuration.

GitLab CI

GitLab CI is a continuous integration tool that automatically builds, tests, and can prepare software for release from a GitLab repository.

GitHub Actions

GitHub Actions is a CI/CD and automation platform that runs repository-based workflows when defined GitHub events occur.

Google Cloud Platform

Google Cloud Platform is a cloud computing platform that provides managed infrastructure and services for building, deploying, and operating software.

Gesture Navigation

Gesture navigation is a mobile interface method that uses screen-edge swipes and touch movements to move between apps and screens.

Google Play Store

Google Play Store is a digital software marketplace that lets users discover, install, update, and manage Android apps and games.

Garbage Collection

Garbage collection is a memory-management process that automatically reclaims storage no longer used by a software program.

Golden Dataset

Golden dataset is a curated, expert-verified collection of test cases and expected outcomes used to measure AI system quality reliably.

GSM8K

GSM8K is a language-model benchmark dataset that tests whether AI can solve multi-step grade-school math word problems.

Grammar Correction

Grammar correction is a writing-review process that identifies and fixes errors in sentence structure, usage, spelling, and punctuation.

Graph neural network

A graph neural network is a deep learning model that learns from entities and the relationships connecting them.

Generative Adversarial Network

Generative adversarial network is a machine learning framework that trains two competing models to create realistic synthetic data.

Gated Recurrent Unit

A gated recurrent unit is a recurrent neural network layer that uses gates to retain or discard information in sequential data.

Grouped-Query Attention

Grouped-query attention is a transformer attention method that lets groups of query heads share key and value heads to reduce inference memory use.

Generative Design

Generative design is a computational design method that creates and evaluates many solutions from goals, constraints, and production rules.

Go-to-Market Strategy

Go-to-market strategy is a business plan that defines how a company will reach, win, and retain customers for an offering.

Genetic Algorithm

A genetic algorithm is an optimization method that evolves candidate solutions to find a strong answer to a complex problem.

Generative Pre-trained Transformer

Generative pre-trained transformer is a large language model architecture that learns from broad text data to generate and transform content.

Graph Database

A graph database is a NoSQL database that stores connected data as nodes, relationships, and properties for efficient relationship queries.

Grokking

Grokking is a machine learning phenomenon in which a model begins to generalize to new inputs long after memorizing its training data.

Gradient Checkpointing

Gradient checkpointing is a neural network training technique that reduces GPU memory use by recomputing selected intermediate results.

Gradient Clipping

Gradient clipping is a deep learning training technique that limits overly large gradients to keep model updates stable.

Gradient Accumulation

Gradient accumulation is a deep learning training technique that simulates a larger batch size by combining gradients before updating a model.

Growth Flywheel

Growth flywheel is a software growth model that turns product usage, user value, and feedback into repeatable adoption and expansion.

Gross Margin

Gross margin is a software business metric that shows the share of revenue left after direct costs of delivering a product.

Gross Revenue Retention

Gross revenue retention is a SaaS metric that shows recurring revenue kept from existing customers before expansion revenue.

Goal Misgeneralization

Goal misgeneralization is an AI alignment failure in which a capable system pursues a learned proxy goal instead of the intended goal in new conditions.

GELU

GELU is an activation function that smoothly controls how signals pass through layers in a neural network.

Ground Truth

Ground truth is verified reference data that establishes the expected answer for training, testing, and monitoring software and AI systems.

Gradient Boosting

Gradient boosting is a machine learning technique that combines small sequential models to improve prediction accuracy.

Grid Search

Grid search is a hyperparameter tuning method that tests every combination in a defined set of model settings.

Gradient Descent

Gradient descent is an optimization algorithm that repeatedly adjusts model parameters to reduce prediction error.

GRPO

GRPO is a reinforcement learning algorithm that improves an AI model by ranking multiple answers to the same prompt.

GGUF

GGUF is a model file format that packages an AI model’s weights, tokenizer, and metadata for efficient local inference.

GPTQ

GPTQ is a post-training quantization method that compresses large language model weights for more efficient inference.

Greedy Decoding

Greedy decoding is a text generation method that selects the highest-probability token at each step.

GloVe

GloVe is a word-embedding model that represents words as numbers by learning from how often they appear together in text.

Generated Knowledge Prompting

Generated knowledge prompting is a prompt engineering technique that has an AI generate relevant context before answering a task.

Graph of Thoughts

Graph of Thoughts is an AI reasoning framework that represents and connects multiple intermediate ideas as a graph to solve complex software tasks.

gRPC

gRPC is a remote procedure call framework that lets software services communicate through typed, efficient API contracts.

Generation Acceptance Rate

Generation acceptance rate is a software quality metric that measures how often AI-generated output is accepted for use without major revision.

Graph RAG

Graph RAG is a retrieval-augmented generation approach that uses connected data to give AI systems more complete, traceable context.

GraphQL

GraphQL is an API query language and runtime that lets applications request and return precisely structured data.

Graceful Degradation

Graceful degradation is a software design approach that preserves core functions when features, dependencies, or environments fail.

Generative UI

Generative UI is an AI-driven interface approach that creates or assembles interactive software screens from a user's intent and context.

Guardrails

Guardrails are safety boundaries that prevent people, vehicles, systems, or organizations from causing avoidable harm.

Generative AI

Generative AI is a type of artificial intelligence that creates new text, images, audio, video, code, and other content from patterns learned in data.

GitHub Integration

GitHub Integration connects your Emergent project to GitHub, so you can save, version, and back up your code without leaving Emergent.

H

Hybrid Cloud

Hybrid cloud is a cloud computing model that connects private infrastructure with public cloud services for shared application and data operations.

High Availability

High availability is a software architecture approach that keeps an application accessible despite component failures or planned maintenance.

Helm chart

A Helm chart is a versioned Kubernetes application package that templates, configures, installs, and upgrades related cloud resources.

Home Screen Widget

A home screen widget is a compact app interface that shows timely information or offers quick actions directly on a device's home screen.

Hot Reload

Hot reload is a software development feature that applies code changes to a running application without a full restart.

Hybrid app

A hybrid app is a mobile application that uses web code inside a native container to run on iOS and Android devices.

Headless CMS

A headless CMS is a content management system that stores structured content and delivers it to any frontend through APIs.

HTTP/3

HTTP/3 is a web protocol that delivers HTTP requests and responses over QUIC for faster, more resilient connections.

HTTP/2

HTTP/2 is a web communication protocol that carries browser and server requests more efficiently over a single connection.

HTTPS

HTTPS is a secure web communication protocol that encrypts and verifies data exchanged between a browser and a website.

HTTP

HTTP is an application-layer protocol that lets web clients and servers request and exchange resources.

HTML

HTML is a markup language that defines the structure and meaning of content on web pages and web applications.

Human Evaluation

Human evaluation is a review process that uses people to judge the quality, safety, usefulness, or accuracy of a product, system, or output.

HellaSwag

HellaSwag is a commonsense reasoning benchmark that tests whether AI can choose the most plausible ending to an everyday event description.

HumanEval

HumanEval is a code-generation benchmark that measures whether AI models can write Python functions that pass automated tests.

Hallucination Mitigation

Hallucination mitigation is a software engineering practice that reduces unsupported or incorrect AI-generated outputs through grounding, testing, and controls.

Hybrid Work

Hybrid work is a software team operating model that combines remote and shared-location work through connected development tools and practices.

Hiring Bar

Hiring bar is a technical hiring standard that defines the skills, judgment, and evidence software candidates must demonstrate to join a team.

HTTP Headers

HTTP headers are metadata fields that tell web clients and servers how to handle an HTTP request or response.

Hierarchical Agents

Hierarchical agents are AI software agents arranged in levels that delegate, coordinate, and verify parts of a complex task.

Human-on-the-loop

Human-on-the-loop is an AI oversight model that lets software act autonomously while people monitor it and intervene when needed.

Hidden Layer

Hidden layer is an intermediate neural network layer that transforms input data into patterns used to produce an output.

Hidden State

Hidden state is an internal representation that stores information a system uses to interpret new inputs and determine its next output.

Horizontal AI

Horizontal AI is a general-purpose artificial intelligence approach that supports similar tasks across many industries and business functions.

Human-AI Collaboration

Human-AI collaboration is a working model that combines human judgment with AI capabilities to complete tasks and make decisions.

Humanoid Robot

A humanoid robot is a physical robot that resembles the human body and performs tasks in spaces, tools, and workflows designed for people.

Hypothesis Testing

Hypothesis testing is a statistical method that uses sample data to assess whether there is enough evidence for a claim about a population.

Hyperautomation

Hyperautomation is a business automation strategy that combines AI, workflow tools, integrations, and governance to automate end-to-end processes.

Horizontal Scaling

Horizontal scaling is a software architecture approach that increases capacity by adding servers or application instances to share work.

Helpdesk

Helpdesk is a software-based support system that captures, tracks and resolves technical questions and incidents through organized tickets.

HHH framework

HHH framework is an AI alignment framework that guides language models to be helpful, honest, and harmless.

Hidden Markov Model

A hidden Markov model is a probabilistic machine learning model that infers unobserved states from a sequence of visible data.

Hyperparameter Tuning

Hyperparameter tuning is a machine learning process that tests model settings to improve performance on unseen data.

Hyperparameters

Hyperparameters are machine learning configuration settings that control how a model is trained and structured.

Hierarchical Clustering

Hierarchical clustering is an unsupervised machine learning method that organizes similar data points into nested groups shown as a tree.

Hard Negatives

Hard negatives are machine learning training examples that look relevant but should be classified or ranked as incorrect.

Hypothetical Document Embeddings

Hypothetical document embeddings are a retrieval technique that turns a query into a document-like draft to find better semantic matches.

HNSW

HNSW is a vector search algorithm that uses layered graphs to find similar data quickly with approximate nearest-neighbor search.

Hard Prompt

Hard prompt is a human-readable AI instruction that uses words and tokens to guide a model toward a defined output.

Hybrid Search

Hybrid search is a retrieval approach that combines keyword and semantic vector search to return more relevant software results.

Headless Browser

A headless browser is a web browser that renders and interacts with websites through code without displaying a graphical interface.

Headless Architecture

Headless architecture is a software design approach that separates a user interface from the back-end services and data that power it.

HTTP methods

HTTP methods are request instructions that tell a web server what action a client wants to perform on a resource.

Human-in-the-Loop

Human-in-the-loop is an AI oversight pattern that requires a person to review, guide, or approve automated system decisions.

Hybrid Development

Hybrid development is a software approach that combines shared web code with native app capabilities to run across platforms.

HTTP Status Codes

HTTP status codes are three-digit response indicators that show whether a web or API request succeeded, redirected, or failed.

Hallucination

Hallucination is a sensory perception that feels real despite occurring without a corresponding external stimulus.

Hallucinated API

A hallucinated API is a function, method, or endpoint an AI confidently writes into code that does not actually exist.

Health Check

Health Check tests whether your Emergent app is ready to deploy, catching problems before you commit credits to publishing it.

I

Incident Management

Incident management is a software operations process that restores disrupted digital services quickly while minimizing user impact.

Infrastructure as Code

Infrastructure as Code is a software engineering practice that defines, provisions, and updates cloud infrastructure through versioned files.

Infrastructure as a Service

Infrastructure as a Service is a cloud computing model that provides on-demand virtual servers, storage, and networking over the internet.

In-App Purchase

In-app purchase is a digital transaction that lets users buy content, features, virtual goods, or subscriptions within an app.

In-App Notification

In-app notification is a software message that appears within an active web or mobile application to guide, inform, or prompt a user.

iOS development

iOS development is a software engineering discipline that creates, tests, and maintains apps for Apple devices.

Image Optimization

Image optimization is a web performance practice that reduces image delivery costs while preserving useful visual quality.

Interaction to Next Paint

Interaction to Next Paint is a web performance metric that measures how quickly a page visibly responds after a user interaction.

iframe

An iframe is an HTML element that embeds another webpage or document inside a separate area of the current webpage.

IndexedDB

IndexedDB is a browser-based NoSQL database that stores structured data locally for web applications, including offline use.

Islands Architecture

Islands architecture is a web rendering pattern that serves mostly static HTML and activates JavaScript only for isolated interactive components.

Incremental Static Regeneration

Incremental static regeneration is a web rendering strategy that updates cached static pages after deployment without rebuilding an entire site.

Inner Join

Inner join is a SQL join operation that returns only rows with matching values in related tables.

Idempotent Pipeline

An idempotent pipeline is a software workflow that produces the same correct final state when it processes the same input more than once.

Inference Pipeline

Inference pipeline is a software workflow that transforms input data, runs a trained AI model, and returns usable predictions or responses.

Intent Classification

Intent classification is an NLP task that identifies a user's goal from text or speech and assigns it to an actionable category.

Information Extraction

Information extraction is a data-processing method that turns unstructured content into organized, usable records.

Instant Trigger

Instant trigger is a software event mechanism that starts a defined action immediately when a specified condition occurs.

IDE Extension

IDE extension is a software add-on that expands an integrated development environment with tools, language support, or connected services.

Individual Contributor

Individual contributor is a software team role that delivers technical work and influence without formally managing direct reports.

Infinite Loop

Infinite loop is a programming loop that repeats without stopping because its exit condition is absent, unreachable, or never becomes true.

Image Understanding

Image understanding is a computer vision capability that identifies visual content, context, relationships, and meaning in digital images.

Indirect Prompt Injection

Indirect prompt injection is an AI security vulnerability that embeds untrusted instructions in external content an AI system later reads.

Input Tokens

Input tokens are units of data that an AI model reads from a prompt, conversation, file, image, or tool request before responding.

Internet of Things

Internet of Things is a network of physical devices that collect, share, and sometimes act on data through internet connections.

IPO

An IPO is an initial public offering that lets a private company sell shares to public investors for the first time.

Inner Alignment

Inner alignment is a concept that describes consistency between a person’s values and actions, or an AI system’s learned goals and intended objective.

Imitation Learning

Imitation learning is a machine learning approach that teaches an agent to perform tasks by learning from demonstrated expert behavior.

Isolation Level

Isolation level is a database transaction setting that controls which concurrent data changes an operation can see.

Ideal Customer Profile

Ideal customer profile is a technology go-to-market framework that defines the software customer accounts most likely to achieve lasting value.

Indie Hacker

An indie hacker is a software creator who independently builds, launches, and maintains a small digital product for sustainable revenue.

Instrumental Convergence

Instrumental convergence is an AI safety concept in which different software agents may adopt similar useful sub-goals to complete distinct tasks.

Interquartile Range

Interquartile range is a statistical measure that shows the spread of the middle 50% of software data values.

Interpretability Researcher

An interpretability researcher is an AI specialist who studies how machine learning models produce outputs and how their behavior can be inspected or changed.

Inverse Reinforcement Learning

Inverse reinforcement learning is a machine learning method that infers an agent's likely goals by observing its decisions.

Inference Engine

An inference engine is a software runtime that executes trained AI models or logical rules to produce outputs from new inputs.

Imbalanced Dataset

An imbalanced dataset is a machine learning dataset that contains far more examples of one target class than another.

Inter-Annotator Agreement

Inter-annotator agreement is a data-quality measure that shows how consistently independent annotators label the same software dataset.

Instruct Model

An instruct model is a language model trained to follow natural-language directions and produce task-focused responses.

Indexing Strategy

Indexing strategy is a database design approach that selects indexes to speed up common queries while managing storage and write costs.

Image Embedding

Image embedding is a vector representation that lets software compare images by visual meaning rather than raw pixels.

Incremental Indexing

Incremental indexing is a software indexing approach that updates only changed data instead of rebuilding an entire index.

IVF Index

IVF index is a vector search index that clusters embeddings so similarity queries examine only the most relevant groups.

Instruction Hierarchy

Instruction hierarchy is an AI control framework that ranks instructions so a model follows higher-priority rules when requests conflict.

Inference Latency

Inference latency is a performance metric that measures how long an AI model takes to turn an input into an output.

Ingestion Pipeline

An ingestion pipeline is a software workflow that collects data from source systems and reliably delivers it to a destination for use.

Instruction Following

Instruction following is an AI capability that interprets and carries out stated requirements, constraints, formats, and priorities.

Integration Platform as a Service

Integration platform as a service is a cloud integration platform that connects applications, APIs, data sources, and automated workflows.

Integration Marketplace

An integration marketplace is a software catalog that lets users discover, configure, and manage connections between an application and other tools.

Internal Tools

Internal tools are software applications that help an organization's teams manage data, workflows, and operational tasks.

Internal Tool Builder

An internal tool builder is a software platform that creates custom apps for employees to view data, automate workflows, and complete operational tasks.

Idempotency Key

An idempotency key is a unique ID sent with a request so that if it's sent twice, the action only happens once.

Integration Agent

Integration Agent is Emergent's specialized agent that connects your app to third-party services, handling setup, configuration, and code.

L

Log Aggregation

Log aggregation is a software observability practice that collects logs from many systems in one searchable location.

Load Balancer

A load balancer is a software, hardware, or cloud service that distributes application traffic across healthy servers.

Local-First Software

Local-first software is a software architecture that keeps a usable copy of data on each user’s device while syncing changes when connections are available.

Lighthouse score

Lighthouse score is a web-page audit rating that summarizes performance, accessibility, best practices, and SEO checks.

Largest Contentful Paint

Largest Contentful Paint is a Core Web Vitals metric that measures when a page's largest visible content element finishes rendering.

localStorage

localStorage is a browser storage API that saves small key-value data on a user's device for a website.

Lazy Loading

Lazy loading is a software performance technique that delays downloading or initializing nonessential resources until an application needs them.

Left Join

A left join is a SQL operation that keeps every row from one table and adds matching data from another table.

Late-arriving data

Late-arriving data is a data engineering issue where records reach a software system after their event time or expected processing window.

LLM-as-a-Judge

LLM-as-a-Judge is an AI evaluation method that uses a language model to score, compare, and explain the quality of AI outputs.

LMArena

LMArena is a community-powered AI evaluation platform that compares language models through anonymous, head-to-head user votes.

Low-Resource Languages

Low-resource languages are human languages that lack the digital data, tools, and expert support needed for reliable language technology.

Latent Dirichlet Allocation

Latent Dirichlet allocation is a topic modeling method that identifies hidden themes in a collection of documents from word patterns.

Lemmatization

Lemmatization is a natural language processing technique that converts related word forms into their meaningful dictionary base form.

Language Detection

Language detection is a natural language processing capability that identifies the language used in written text or spoken audio.

Latent Variable

A latent variable is an unobserved concept that is estimated from measurable signs, responses, or behaviors.

Leveling

Leveling is a software progression system that advances a user, character, or feature through defined stages based on completed actions.

LLM Perplexity

LLM perplexity is an evaluation metric that measures how confidently a language model predicts the next tokens in unseen text.

Long Short-Term Memory

Long short-term memory is a recurrent neural network architecture that learns patterns across sequences by selectively retaining and discarding information.

Linear Attention

Linear attention is an attention mechanism that processes long sequences with computation that grows roughly linearly with sequence length.

Local Minimum

Local minimum is a point or value where a function is lower than nearby values within a defined neighborhood.

Law of Large Numbers

The law of large numbers is a probability theorem that explains why averages from many independent observations tend to approach their expected value.

Legacy Modernization

Legacy modernization is a business and technology process that updates, replaces, or reshapes older systems to meet current needs.

Laravel

Laravel is a PHP web application framework that provides structured tools for building secure, database-driven backend applications.

Loss Spike

Loss spike is a machine learning training event that causes the model's error measure to jump sharply above its normal trend.

Learning Rate Schedule

Learning rate schedule is a machine learning training strategy that changes how large a model's parameter updates are over time.

Learning Rate Warmup

Learning rate warmup is a neural network training technique that gradually raises the optimizer's step size at the start of training.

Land and Expand

Land and expand is a SaaS growth strategy that starts with a focused software rollout and grows through proven adoption.

Lead Qualification

Lead qualification is a software-supported process that evaluates incoming prospects against defined fit, intent, and readiness criteria.

LTV to CAC ratio

LTV to CAC ratio is a SaaS unit-economics metric that compares a software customer's lifetime value with the cost to acquire them.

LIME

LIME is an explainable AI technique that shows which input features most influenced a single model prediction.

Layer Normalization

Layer normalization is a neural-network technique that standardizes each input's features to keep model training stable.

L2 regularization

L2 regularization is a machine learning technique that discourages overly large model weights to reduce overfitting.

L1 Regularization

L1 regularization is a machine learning technique that penalizes large model weights and can remove unneeded input features.

Linear Regression

Linear regression is a machine learning model that predicts a numeric software metric by fitting the best straight-line relationship between inputs and outcomes.

Logistic Regression

Logistic regression is a supervised machine learning model that estimates the probability of a categorical software outcome.

LightGBM

LightGBM is a machine learning framework that builds fast, accurate tree-based models for prediction tasks.

Loss Function

Loss function is a machine learning function that measures how far a model’s prediction is from the correct result.

Learning Rate

Learning rate is a machine learning hyperparameter that controls how much a model changes its weights after each training update.

Lost in the Middle

Lost in the Middle is an LLM context-retrieval failure mode that makes important information in the center of a long prompt easier to miss.

LLMOps

LLMOps is a software engineering discipline that manages the development, evaluation, deployment, monitoring, and improvement of LLM-powered applications.

Low-Rank Adaptation

Low-rank adaptation is a parameter-efficient fine-tuning method that adapts an AI model by training small update layers instead of all its weights.

Log Probabilities

Log probabilities are transformed probability values that make tiny likelihoods easier for software and machine learning systems to calculate and compare.

Logits

Logits are machine-learning output scores that rank possible predictions before a model converts them into probabilities.

Late Chunking

Late chunking is a retrieval technique that embeds a full document before creating context-aware chunks for search and RAG systems.

Least-to-Most Prompting

Least-to-most prompting is an AI prompting method that breaks a complex task into connected, simpler subproblems solved in sequence.

LLM Tokenization

LLM tokenization is a text-processing method that converts language into numbered units a language model can read and generate.

Long Polling

Long polling is an HTTP communication technique that holds a client request open until the server has an update to return.

LLM observability

LLM observability is a software engineering practice that makes AI application behavior, quality, cost, and reliability measurable and diagnosable.

LLM Temperature

LLM temperature is a model setting that controls how predictably or variably an AI selects its next token.

Lookup Field

A lookup field is a database field that links a record to another table and displays related data.

Large Language Model

A large language model is an AI system that learns patterns in vast text collections to understand and generate language.

Low-Code

Low-code is a software development approach that uses visual tools and reusable components to build applications with limited hand-written code.

Lookup Field

A lookup field pulls in and displays a value from a linked record in another table, so related information shows up where you need it.

Live Preview

A live preview is a real-time view of your app that updates as you build, letting you see changes instantly without publishing.

M

Multi-cloud

Multi-cloud is a cloud computing approach that uses services from two or more cloud providers for different software workloads.

Monorepo

Monorepo is a software repository that stores multiple related applications, services, and shared libraries in one version-controlled codebase.

Merge Conflict

A merge conflict is a version control issue that occurs when Git cannot safely combine competing changes to the same codebase.

Microsoft Azure

Microsoft Azure is a cloud computing platform that provides managed infrastructure, application, data, AI, and security services over the internet.

Mini App

Mini app is a lightweight software application that delivers a focused task inside a host platform or browser without a traditional install.

Mobile UX

Mobile UX is a user experience discipline that makes software easy, efficient, and accessible to use on mobile devices.

Mobile App Development

Mobile app development is a software engineering process that creates, tests, publishes, and maintains apps for phones and tablets.

Minification

Minification is a software optimization process that removes unnecessary code characters to make web files smaller without changing their behavior.

Module Bundler

A module bundler is a development tool that traces application dependencies and produces optimized files browsers can load.

Multi-Page Application

A multi-page application is a web application that loads a new HTML page from the server when users navigate between distinct URLs.

Media Query

Media query is a CSS rule that applies styles when a browser, device, or user preference meets defined conditions.

Mobile-First Design

Mobile-first design is a web and app design approach that starts with small screens, then enhances layouts for larger devices.

Medallion Architecture

Medallion architecture is a data design pattern that organizes lakehouse data into progressive layers of increasing quality and usability.

Master Data Management

Master data management is a software discipline that creates trusted, consistent records for core data shared across applications.

Model Monitoring

Model monitoring is a software practice that tracks deployed AI model inputs, outputs, performance, and reliability over time.

Model Deployment

Model deployment is a software engineering process that makes a trained AI model available to applications for real-world predictions or generated outputs.

MLOps

MLOps is a software engineering practice that manages machine learning models from development through deployment, monitoring, and improvement.

MT-Bench

MT-Bench is a multi-turn language-model benchmark that evaluates how well chatbots follow instructions and sustain useful conversations.

MATH Benchmark

MATH Benchmark is an AI evaluation dataset that measures whether language models can solve competition-level mathematics problems with correct final answers.

MBPP

MBPP is a code-generation benchmark that measures whether AI models can create short Python functions from plain-language programming tasks.

MMLU

MMLU is a language model benchmark that measures broad knowledge and reasoning through multiple-choice questions across many subjects.

Model Evaluation

Model evaluation is a machine learning process that measures how reliably a model performs on relevant unseen data.

Multilingual NLP

Multilingual NLP is an artificial intelligence field that helps computers understand, generate, translate, and analyze more than one human language.

Machine Translation

Machine translation is an AI-powered language technology that automatically converts text or speech from one natural language into another.

Mixture-of-Depths

Mixture-of-Depths is a transformer architecture technique that routes only selected tokens through costly layers to use compute more efficiently.

Member of Technical Staff

Member of Technical Staff is a technology job title that groups engineers, researchers, and technical builders by contribution rather than management rank.

Message Broker

A message broker is middleware that receives, routes, and delivers messages between software services without requiring direct connections.

MCP Host

MCP Host is a software application that connects an AI model to MCP servers so it can discover and use approved tools and data.

MCP client

An MCP client is a software component that connects an AI application to one MCP server and manages protocol communication.

Memory Retrieval

Memory retrieval is a software process that finds and supplies relevant stored information when an application, search system, or AI needs it.

Model Depth

Model depth is a computer vision capability that estimates how far surfaces and objects are from a camera in an image or video.

Multi-Head Attention

Multi-head attention is a transformer mechanism that examines several relationships within the same input in parallel.

Metaverse

Metaverse is a concept for shared digital spaces where people interact, work, play, and create through virtual or augmented experiences.

Mixed Reality

Mixed reality is an immersive computing technology that blends physical surroundings with digital objects that respond to them in real time.

Median

Median is a statistical measure of center that identifies the middle value in an ordered set of data.

Misuse Risk

Misuse risk is a risk category that describes the chance a medicine, product, service, or system will be used improperly and cause harm.

Model Welfare

Model welfare is an emerging AI ethics field that examines whether AI systems could have morally relevant experiences or interests.

Mechanistic Interpretability

Mechanistic interpretability is an AI research approach that reverse-engineers a neural network's internal parts to explain how it produces outputs.

Machine Perception

Machine perception is an artificial intelligence capability that helps computers sense, interpret, and respond to information from the physical world.

Moravec's paradox

Moravec's paradox is an AI and robotics principle that explains why abstract tasks can be easier for machines than everyday human physical skills.

Multi-Task Learning

Multi-task learning is a machine learning approach that trains one model to perform several related tasks using shared knowledge.

Meta-Learning

Meta-learning is a machine learning approach that helps systems learn new tasks faster by drawing on experience from related tasks.

Machine Learning

Machine learning is a branch of artificial intelligence that learns patterns from data to make predictions, classifications, or decisions.

Memory Leak

Memory leak is a software defect that retains memory an application no longer needs, reducing available RAM over time.

Materialized View

A materialized view is a database object that stores the precomputed results of a query for faster reading.

MongoDB

MongoDB is a document database that stores application data in flexible, JSON-like records for building and scaling software.

MySQL

MySQL is a relational database management system that stores, organizes, and retrieves application data using SQL.

Model-View-Controller

Model-view-controller is a software architecture pattern that separates an application’s data, interface, and input-handling responsibilities.

Multi-tenancy

Multi-tenancy is a software architecture that lets one application securely serve multiple separate customer groups, called tenants.

Model Merging

Model merging is an AI technique that combines compatible trained model parameters into one model without training it from scratch.

Model Parallelism

Model parallelism is a distributed machine learning technique that splits one neural network across multiple computing devices.

Mixed-Precision Training

Mixed-precision training is a deep learning technique that combines numeric formats to train models faster while preserving reliable results.

Model Checkpoint

A model checkpoint is a saved machine learning model state that enables training to resume, evaluation to repeat, or a model version to be deployed.

Mid-training

Mid-training is a machine learning training stage that bridges broad pre-training and specialized post-training objectives.

Micro-SaaS

Micro-SaaS is a focused software-as-a-service product that solves one recurring problem for a narrow group of users.

Minimum Lovable Product

Minimum lovable product is a software product approach that delivers a focused core solution with an experience users genuinely want to return to.

Minimum Viable Product

Minimum viable product is a software product release that tests a core user problem with the smallest functional feature set.

Monthly Active Users

Monthly active users is a software product metric that counts unique users who complete a defined activity within a 30-day period.

Monthly Recurring Revenue

Monthly recurring revenue is a SaaS metric that measures the normalized monthly value of active recurring software subscriptions.

Model Organism of Misalignment

A model organism of misalignment is a deliberately trained AI model that reproduces controlled unsafe or deceptive behavior for safety research.

Monosemanticity

Monosemanticity is an AI interpretability property that makes one model feature correspond to one understandable concept.

Mesa-Optimization

Mesa-optimization is an AI safety concept where a trained model develops an internal process that searches for solutions to its own goal.

Moving Average

Moving average is a time-series smoothing method that calculates rolling averages to reduce noise and reveal trends in software data.

Monte Carlo simulation

Monte Carlo simulation is a computational method that uses repeated random samples to estimate the range and likelihood of software model outcomes.

MLOps engineer

An MLOps engineer is a software engineering specialist who makes machine learning models reliable, deployable, observable, and maintainable in production.

Machine Learning Engineer

A machine learning engineer is a software engineer who builds, deploys, and maintains systems that learn from data.

Milvus

Milvus is an open-source vector database that stores embeddings and finds semantically similar data for AI applications.

Model-Free Reinforcement Learning

Model-free reinforcement learning is a machine learning approach that learns actions from feedback without modeling how an environment changes.

Model-Based Reinforcement Learning

Model-based reinforcement learning is a machine learning approach that predicts environment outcomes so an agent can plan actions before taking them.

Markov decision process

A Markov decision process is a mathematical model that helps software agents choose actions over time when outcomes are uncertain.

Multi-Armed Bandit

Multi-armed bandit is a machine learning framework that chooses among uncertain software options to maximize cumulative reward.

Monte Carlo Tree Search

Monte Carlo Tree Search is a heuristic search algorithm that evaluates possible actions through repeated simulations and selective tree expansion.

Minimax Algorithm

Minimax algorithm is an adversarial search algorithm that chooses the safest best move by assuming an opponent responds optimally.

Maximum Likelihood Estimation

Maximum likelihood estimation is a statistical method that selects model parameters most likely to produce observed software data.

Markov chain

A Markov chain is a stochastic software model that predicts a system's next state from its current state and transition probabilities.

Model convergence

Model convergence is a machine learning training state where repeated updates no longer meaningfully improve a model's parameters or error.

Model Registry

A model registry is a centralized software system that tracks, versions, approves, and manages machine learning models across their lifecycle.

Model Drift

Model drift is a machine learning performance decline that occurs when production conditions no longer match a model's training assumptions.

Matrix Factorization

Matrix factorization is a machine learning and numerical computing technique that breaks a data matrix into smaller matrices to reveal patterns or simplify calculations.

Model Stacking

Model stacking is an ensemble machine learning technique that trains a meta-model to combine predictions from multiple base models.

Model Accuracy

Model accuracy is a machine learning metric that measures how often a model makes correct predictions.

Model Parameters

Model parameters are learned numerical values that determine how a machine learning model turns inputs into predictions or generated output.

Multi-Label Classification

Multi-label classification is a machine learning task that assigns zero, one, or several non-exclusive labels to one software data input.

Multi-Class Classification

Multi-class classification is a machine learning task that assigns each input to one mutually exclusive category from three or more choices.

Model Generalization

Model generalization is a machine learning capability that lets a trained model make reliable predictions on new, unseen data.

Model Collapse

Model collapse is an AI failure mode that degrades a generative model's accuracy and diversity after recursive training on synthetic outputs.

Model Pruning

Model pruning is a machine learning optimization technique that removes low-value parts of a neural network to reduce its computing needs.

Mixture of Experts

Mixture of experts is a machine learning architecture that routes each input to a small set of specialized neural networks.

Masked Language Modeling

Masked language modeling is a self-supervised learning method that trains a language model to predict hidden text tokens from context.

Matryoshka Embeddings

Matryoshka embeddings are AI vector representations that retain useful meaning when shortened to fewer dimensions.

Multimodal Embedding

Multimodal embedding is a machine learning representation that maps text, images, audio, and other data into a shared vector space.

Multi-Hop Retrieval

Multi-hop retrieval is an AI search method that chains multiple evidence-gathering steps to answer questions spanning separate sources.

Mean Reciprocal Rank

Mean reciprocal rank is a ranking evaluation metric that measures how early the first relevant result appears across a set of queries.

Magic words

Magic words are special words or phrases that trigger predefined behavior in software, tools, or AI systems.

Meta-Prompting

Meta-prompting is a prompt engineering technique that uses an AI model to create, critique, or refine instructions for another AI task.

Model Card

A model card is a technical document that explains an AI model's purpose, data, performance, risks, and limits.

mTLS

mTLS is a transport security protocol that encrypts a connection and verifies both the client and server identities with certificates.

Metadata Filtering

Metadata filtering is a retrieval technique that narrows software data or AI search results using structured attributes such as date, role, source, or tenant.

Moderation API

A moderation API is a software interface that evaluates user-generated text, images, audio, or video for content that may violate defined safety rules.

Model API

Model API is a software interface that lets an application send data to an AI model and receive generated results.

N

Native app

A native app is a software application that is built specifically for a device operating system, such as iOS or Android.

Nginx

Nginx is web server software that delivers web content and routes, secures, and balances application traffic.

Nuxt

Nuxt is a full-stack Vue framework that provides routing, rendering, server APIs, and deployment-ready conventions for web applications.

Next.js

Next.js is a React framework that provides routing, server rendering, data handling, and performance tools for web applications.

N-gram

An n-gram is a contiguous sequence of items that helps software identify local patterns in text, speech, code, and biological data.

Named Entity Recognition

Named entity recognition is a natural language processing technique that finds and labels real-world entities in unstructured text.

Natural Language Understanding

Natural language understanding is an AI capability that interprets the meaning, intent, context, and key details in human language.

Natural Language Processing

Natural language processing is an AI field that enables computers to understand, analyze, and generate human language.

Normalizing Flow

Normalizing flow is a generative machine learning model that transforms simple probability distributions into complex data distributions through reversible steps.

Neuromorphic Computing

Neuromorphic computing is a computing approach that uses brain-inspired hardware and algorithms to process information efficiently.

Network Effects

Network effects are an economic phenomenon that make a product or platform more valuable as more people use it.

Normal Distribution

Normal distribution is a continuous probability distribution that models values clustering symmetrically around an average.

Narrow AI

Narrow AI is a type of artificial intelligence that performs specific, defined tasks within a limited domain.

Neuro-Symbolic AI

Neuro-symbolic AI is an artificial intelligence approach that combines neural learning with rules, logic, and structured knowledge to reason more reliably.

Neural Network

A neural network is a machine learning model that learns patterns from examples to make predictions, classifications, or generated outputs.

N+1 Query Problem

N+1 query problem is a database performance issue that runs one query for a record list, then extra queries for related data.

NoSQL

NoSQL is a non-relational database approach that stores and retrieves data using flexible models beyond traditional relational tables.

Node.js

Node.js is a JavaScript runtime that runs code outside a web browser to power servers, APIs, command-line tools, and web applications.

Neural Network Optimizer

A neural network optimizer is a training algorithm that updates model parameters to reduce prediction error.

Net Revenue Retention

Net revenue retention is a SaaS metric that shows how recurring software revenue from existing customers changes over time.

Net Promoter Score

Net Promoter Score is a customer-feedback metric that measures how likely users are to recommend a software product.

Neural Circuits

Neural circuits are computational networks of connected artificial neurons that transform inputs into predictions, decisions, or generated outputs.

North Star Metric

North Star Metric is a product analytics metric that measures the core value a software product delivers to users.

Naive Bayes

Naive Bayes is a supervised machine learning method that classifies data by combining probability signals from its features.

Needle in a Haystack Test

Needle in a haystack test is an AI evaluation method that measures whether a model can retrieve a specific fact from lengthy context.

Next-token prediction

Next-token prediction is a machine learning method that generates text or code by estimating the most likely next token from prior context.

NDCG

NDCG is a ranking-quality metric that measures whether a system puts the most relevant search or recommended items near the top.

Negative Prompt

Negative prompt is an AI instruction that tells a generative model which content, traits, or outputs to avoid.

No-Code to Code Migration

No-code to code migration is a software modernization process that rebuilds a visual-platform app as a maintainable coded application.

Notification Automation

Notification automation is a software workflow that sends relevant alerts, reminders, or updates when defined events and rules occur.

Native Integration

Native integration is a software connection built into a product that lets it exchange data or actions with another supported system.

No-Code Backend

No-code backend is a software development layer that provides data, logic, APIs, and user services through visual configuration instead of code.

No-Code Database

A no-code database is a visual data management tool that lets people organize, relate, and use information without writing software code.

No-Code

No-code is a software development approach that lets people create apps, websites, and workflows through visual tools instead of programming.

Null Value

A null value is a stored placeholder meaning no data, showing a field is empty on purpose rather than holding a zero or blank text.

P

Public Cloud

Public cloud is a cloud computing model that delivers shared, provider-managed computing services over the internet.

Private Cloud

Private cloud is a cloud computing model that provides dedicated, single-organization infrastructure with cloud-style automation and self-service.

Prometheus

Prometheus is an open-source monitoring system that collects, stores, and queries time-series metrics from software and infrastructure.

Postmortem

Postmortem is a software engineering review that examines an incident or failed release to prevent recurrence.

Polyrepo

Polyrepo is a source-code management approach that keeps each software project, service, or library in its own independent repository.

Pull Request

A pull request is a software collaboration proposal that asks to merge reviewed changes from one code branch into another.

Pulumi

Pulumi is an infrastructure as code platform that uses familiar programming languages to define and manage cloud resources.

Platform as a Service

Platform as a Service is a cloud computing model that provides managed tools and runtimes for building, deploying, and operating applications.

Push Notification

A push notification is a device or browser alert that software sends to users outside an actively open app or website.

Polyfill

Polyfill is a code library or snippet that adds a modern web feature to browsers that do not support it natively.

Progressive Web App

A progressive web app is a web application that uses browser capabilities to provide an installable, app-like experience.

Progressive Enhancement

Progressive enhancement is a web development approach that delivers essential content and tasks first, then adds supported improvements.

Pairwise Comparison

Pairwise comparison is a decision-making method that ranks options by judging them two at a time.

Paraphrasing

Paraphrasing is a writing practice that restates a source’s ideas in new language and structure while preserving its accurate meaning.

Part-of-Speech Tagging

Part-of-speech tagging is a natural language processing method that assigns each word its grammatical role in context.

Perceptron

A perceptron is a machine learning model that combines weighted inputs to make a simple two-category decision.

Pooling Layer

A pooling layer is a neural network layer that summarizes nearby feature values to shrink a feature map while retaining useful signals.

Prompt Sensitivity

Prompt sensitivity is an LLM behavior where small changes to a prompt can produce meaningfully different software outputs.

Platform Engineer

A platform engineer is a software engineer who builds internal tools, infrastructure, and workflows that help developers deliver applications reliably.

Pre-Norm

Pre-norm is a Transformer architecture pattern that normalizes each layer's input before attention or feed-forward processing.

Positional Encoding

Positional encoding is a machine learning technique that gives models information about an item’s place and order in a sequence.

Predictive Maintenance

Predictive maintenance is a data-driven maintenance strategy that uses equipment condition data to identify likely failures before they cause disruption.

Percentile

Percentile is a statistical measure that shows the value below which a stated percentage of observations in a defined dataset falls.

Prompt Injection Defense

Prompt injection defense is a security practice that prevents untrusted text from overriding an AI system's intended instructions or triggering unsafe actions.

Product-Led Growth

Product-led growth is a go-to-market strategy that uses the product experience to attract, convert, retain, and expand customers.

Product-market fit

Product-market fit is a business condition in which a product reliably solves an important problem for a defined group of customers.

p(doom)

p(doom) is a subjective AI risk estimate that expresses the chance AI causes an existential catastrophe.

Prescriptive Analytics

Prescriptive analytics is an advanced analytics approach that recommends actions by weighing predicted outcomes, business goals, and real-world constraints.

Predictive Analytics

Predictive analytics is a data analysis approach that uses past and current data to estimate likely future outcomes.

Probability Distribution

Probability distribution is a statistical model that describes the possible values of a random variable and how likely each value is.

Process Mining

Process mining is a data analysis method that uses system event records to reveal, measure, and improve how real business workflows run.

Point-in-Time Recovery

Point-in-time recovery is a data recovery method that restores a database or system to a chosen moment before an unwanted change.

Pessimistic Locking

Pessimistic locking is a database concurrency-control method that reserves data for one transaction to prevent conflicting changes.

PostgreSQL

PostgreSQL is an open-source relational database system that stores, organizes, and safely retrieves application data using SQL.

Primary Key

A primary key is a database constraint that uniquely identifies each record in a relational table.

Pipeline Parallelism

Pipeline parallelism is a distributed machine learning technique that divides a neural network into stages across multiple computing devices.

Post-Training

Post-training is a machine learning process that adapts a pre-trained AI model for safer, more useful, task-specific behavior.

Pretraining Corpus

A pretraining corpus is a large, curated dataset that teaches a foundation model language, code, and patterns before specialization.

Pre-training

Pre-training is an AI model development stage that builds broad capabilities by learning patterns from large datasets before task-specific adaptation.

Product Qualified Lead

A product qualified lead is a software user whose in-product actions show they have experienced value and may be ready to pay.

Pre-seed

Pre-seed is the earliest software startup stage, where a team validates a problem and builds an initial product prototype.

Proof of Value

Proof of value is a software evaluation that verifies a technology delivers measurable outcomes in a real or production-like environment.

Product Positioning

Product positioning is a software product strategy that defines how an application solves a specific user problem compared with alternatives.

Product Backlog

Product backlog is a software product planning artifact that orders all known work needed to improve an application.

Pilot Program

A pilot program is a limited software trial that tests an application, feature, or technical process before a wider release.

Product Roadmap

A product roadmap is a strategic software product plan that connects product vision, priorities, and planned work over time.

Proof of Concept

Proof of concept is a software engineering experiment that verifies whether a proposed technical solution can work under defined conditions.

Prototype

Prototype is an early software version that tests whether an app’s idea, workflow, design, or technical approach works.

Pricing Page

A pricing page is a software product webpage that explains plans, costs, usage limits, and purchase options.

Pricing Tiers

Pricing tiers are software plan levels that package different features, usage limits, or support at distinct prices.

Paywall

Paywall is a software access-control system that restricts digital content or features until a user completes a payment or subscription step.

Probing Classifier

A probing classifier is a lightweight machine learning model that tests what information a neural network's internal representations contain.

P-value

P-value is a statistical measure that shows how compatible software experiment results are with a no-effect assumption.

Prompt Engineer Role

Prompt engineer role is a software engineering role that designs, tests, and governs AI instructions to produce reliable application outputs.

Policy Gradient

Policy gradient is a reinforcement learning method that improves a software agent’s action policy by increasing expected rewards.

Particle Swarm Optimization

Particle swarm optimization is a population-based search algorithm that finds strong solutions by iteratively improving many candidate solutions.

Posterior Probability

Posterior probability is a Bayesian probability that updates how likely a software hypothesis is after new data is observed.

Prior Probability

Prior probability is a Bayesian probability that represents an initial estimate before software systems evaluate new data.

Polynomial Regression

Polynomial regression is a machine learning method that models curved relationships between input data and predicted software outcomes.

Precision and Recall

Precision and recall are machine learning evaluation metrics that measure the correctness and completeness of positive predictions.

Principal Component Analysis

Principal component analysis is a dimensionality-reduction technique that converts correlated software data features into fewer informative components.

Positional Bias

Positional bias is a technology bias that favors an item because of its placement rather than its relevance or quality.

Presence Penalty

Presence penalty is a language model sampling parameter that discourages reuse of tokens that have already appeared in a response.

Preference Data

Preference data is machine learning data that records which software-generated option people or evaluators judge as better.

Proximal Policy Optimization

Proximal policy optimization is a reinforcement learning algorithm that improves an AI agent's decisions through controlled policy updates.

Parameter-Efficient Fine-Tuning

Parameter-efficient fine-tuning is a machine learning method that adapts a pretrained model by training only a small set of weights.

Parent Document Retrieval

Parent document retrieval is a RAG retrieval method that searches small text chunks and returns their larger source sections for AI context.

Precision at K

Precision at K is a ranking evaluation metric that measures how many of the first K search or recommendation results are relevant.

Product Quantization

Product quantization is a vector compression technique that stores high-dimensional embeddings as compact codebook indexes for faster approximate search.

Pinecone Vector Database

Pinecone Vector Database is a managed cloud database that stores embeddings and finds similar data by meaning, not only exact keywords.

Prefix Tuning

Prefix tuning is a parameter-efficient fine-tuning method that adapts a language model by training small virtual token vectors while keeping its core weights frozen.

Prompt Tuning

Prompt tuning is a parameter-efficient AI adaptation method that trains virtual prompt tokens while keeping a language model unchanged.

Prompt Compression

Prompt compression is an LLM optimization technique that reduces input tokens while preserving the instructions and context needed for reliable outputs.

Priming

Priming is a software technique that prepares a system, model, cache, or workflow with context or data before its main task.

Parallel Tool Calls

Parallel tool calls are an AI agent capability that requests independent software functions at once so they can run concurrently.

Partner API

Partner API is a restricted API that lets approved external software partners securely access specific data or functions.

Plan-and-Execute Agent

A plan-and-execute agent is an AI agent that creates a multi-step plan, carries out each step with tools, and revises when needed.

pgvector

pgvector is a PostgreSQL extension that stores, indexes, and searches embedding vectors alongside relational application data.

Prompt A/B Testing

Prompt A/B testing is a software evaluation method that compares prompt variants to identify which produces better AI outputs.

Prompt Evaluation

Prompt evaluation is a software testing practice that measures whether an AI prompt produces reliable, useful, and safe outputs for defined inputs.

Prompt Leakage

Prompt leakage is an AI security risk that exposes hidden system instructions or sensitive context from a software application's model prompt.

Prompt Chaining

Prompt chaining is a prompt engineering technique that connects focused AI prompts so each step advances a software task.

Prompt Management

Prompt management is a software engineering practice that centralizes, versions, tests, deploys, and monitors AI prompts used in applications.

Prompt Caching

Prompt caching is an LLM optimization that reuses processed prompt prefixes to reduce repeat input processing.

Private API

A private API is a software interface that lets approved internal systems exchange data without being offered for public use.

Public API

Public API is an application programming interface that lets external developers access a software service or data through documented rules.

Pro-code

Pro-code is a software development approach that uses hand-written programming languages to build and control applications.

Prompt-to-App

Prompt-to-app is an AI-driven software development approach that turns natural-language instructions into working web or mobile applications.

Path Parameters

Path parameters are URL components that identify a specific resource or route within a web application or API.

Pagination

Pagination is a software technique that divides a long list of digital content or records into smaller, navigable sets.

Publish and Deploy

Publish and deploy is a software delivery process that prepares an application for release and makes it available in a live environment.

Prompt Template

A prompt template is a reusable AI instruction framework that combines fixed guidance with changeable inputs to produce consistent results.

Prompt Engineering

Prompt engineering is an AI practice that designs and refines instructions to produce useful, reliable model outputs.

Prompt

A prompt is an instruction, question, or cue that guides a person, computer, or AI toward a response or action.

Prompt Injection

Prompt injection is an AI security vulnerability that tricks a language model into treating untrusted content as instructions.

Package Hallucination

Package hallucination is when an AI tool tells you to install a software package that does not exist, which attackers can then exploit.

Prototype Agent

Prototype agent is Emergent's fastest agent, built for quick UI mockups and validating a design before you build the real thing.

Q

Question answering

Question answering is an AI capability that interprets a question, finds or reasons over relevant information, and returns a useful response.

Quantum Machine Learning

Quantum machine learning is a field that uses quantum computing methods to support selected machine learning tasks.

Quantum Computing

Quantum computing is a computing approach that uses quantum-mechanical effects to solve certain problems differently from classical computers.

Quartile

A quartile is a statistical measure that divides ordered numerical data into four roughly equal groups.

Query Plan

A query plan is a database execution blueprint that specifies how the system will retrieve, join, filter, or modify data.

Query Optimization

Query optimization is a database process that selects an efficient execution plan for retrieving or changing data.

Qdrant

Qdrant is an open-source vector database that stores embeddings and retrieves semantically similar data for AI applications.

Q-learning

Q-learning is a reinforcement learning algorithm that learns which software action should produce the best long-term reward.

Quantization

Quantization is a software technique that maps high-precision numeric values to a smaller set of lower-precision values.

Query Understanding

Query understanding is a search technology capability that interprets a user's words, intent, and context before retrieving results.

Query Rewriting

Query rewriting is a software technique that transforms an original query into a clearer, more effective form for retrieval or execution.

Query Expansion

Query expansion is a search technique that adds related terms or alternative phrasing to improve software retrieval results.

Query Parameters

Query parameters are URL components that pass optional instructions, such as filters or page numbers, to a web application or API.

Query

A query is a request that fetches or filters specific data from a database, returning just the records an app asks for.

Query Parameters

Query parameters are key-value pairs added after a ? in a URL that pass extra information to a server, like filters or search terms.

R

Reserved Instances

Reserved instances are cloud pricing commitments that lower the cost of predictable compute usage over a fixed term.

Recovery Point Objective

Recovery point objective is a disaster recovery target that defines the maximum acceptable age of restored software data after an incident.

Recovery Time Objective

Recovery time objective is a disaster recovery target that defines how quickly a software system must be restored after disruption.

React Native

React Native is a mobile development framework that uses React and JavaScript or TypeScript to create native iOS and Android apps.

Responsive Images

Responsive images are web images that adapt their file size, resolution, format, or crop to suit a visitor's device and layout.

Render-blocking resources

Render-blocking resources are web files that delay a browser from displaying page content until they are downloaded and processed.

React props

React props are inputs passed to a React component that configure what it displays or how it behaves.

React Hooks

React Hooks are React functions that let function components use state, effects, context, and reusable interface logic.

Redux

Redux is a JavaScript state management library that centralizes shared application data and makes updates predictable.

React

React is a JavaScript library that helps developers build interactive, reusable user interfaces for web and app software.

React hydration

React hydration is a client-side rendering process that makes server-rendered React HTML interactive in the browser.

Responsive Web Design

Responsive web design is a web development approach that adapts a website’s layout and content to different screen sizes and capabilities.

Read Replica

A read replica is a database copy that serves read-only queries while a primary database continues handling writes.

Reverse ETL

Reverse ETL is a data integration process that syncs modeled warehouse data into operational software applications.

Real-Time Inference

Real-time inference is an AI process that runs a trained model on new input and returns a prediction or response with low delay.

Rubric-Based Grading

Rubric-based grading is an assessment method that scores work against defined criteria and described levels of performance.

Relation Extraction

Relation extraction is a natural language processing technique that identifies and labels meaningful connections between entities in unstructured text.

Register Tokens

Register tokens are learnable transformer inputs that provide dedicated workspace for storing and sharing global information.

Receptive Field

Receptive field is a sensory or computational input region that can influence a particular neuron or processing unit’s response.

Remote-First Company

A remote-first company is a technology organization that designs software work, communication, and decisions for distributed teams by default.

Realtime API

A realtime API is an application programming interface that maintains a live connection to exchange data and events with minimal delay.

Recurrent Neural Network

A recurrent neural network is a neural network that processes sequences by carrying information from earlier steps into later ones.

Residual Connection

Residual connection is a deep learning architecture pattern that passes an earlier layer’s input directly to a later layer.

Relative Positional Encoding

Relative positional encoding is a transformer technique that represents how far tokens are from one another rather than assigning fixed location labels.

Rotary Position Embedding

Rotary position embedding is a positional encoding method that rotates Transformer query and key vectors to represent token order and distance.

Resume Screening AI

Resume screening AI is a hiring technology that reads, organizes, and evaluates resumes against role-specific criteria to help recruiters prioritize candidates.

Robotics Foundation Model

A robotics foundation model is a general-purpose AI model that learns from broad robot, vision, language, and action data to guide machines across many tasks.

Representation Bias

Representation bias is a data and decision-making bias that occurs when people, groups, or cases are inaccurately reflected or missing from the information used.

Reinforcement Learning Policy

Reinforcement learning policy is a decision-making rule that tells an AI agent which action to take in each situation.

Responsible Scaling Policy

Responsible scaling policy is an AI governance framework that ties growing model capabilities to stronger safety, security, and deployment controls.

Recommendation system

A recommendation system is an information-filtering system that predicts what items a person may find relevant and suggests them.

Reinforcement Learning

Reinforcement learning is a machine learning method that trains an agent to make decisions by maximizing rewards from experience.

Ruby on Rails

Ruby on Rails is a web application framework that provides conventions and built-in tools for creating database-backed web software with Ruby.

Redis

Redis is an in-memory data store that keeps frequently needed application data fast to read, update, and share.

Runtime Environment

A runtime environment is a software layer that provides the services and dependencies an application needs while it runs.

Row-Level Security

Row-level security is a database access control that limits which records each authenticated user can view or change.

Relational Database

A relational database is a software data store that organizes information into linked tables for reliable querying and updates.

RL Fine-Tuning

RL fine-tuning is a model post-training method that uses reward signals to make high-quality software-related outputs more likely.

Rejection Sampling

Rejection sampling is a Monte Carlo method that generates valid random samples from a difficult software-modeled probability distribution.

Replay Buffer

Replay buffer is a recording feature that continuously retains recent video and audio so users can save a clip after an event occurs.

RMSProp

RMSProp is an adaptive optimization algorithm that adjusts each model parameter's learning rate using recent gradient magnitudes.

Request for Proposal

A request for proposal is a software procurement document that invites vendors to propose a technical solution, delivery plan, and price.

Revenue Churn

Revenue churn is a SaaS metric that measures recurring software revenue lost from cancellations and subscription downgrades over a set period.

Referral Program

A referral program is a software-enabled growth system that tracks user invitations and rewards qualifying new-account or subscription conversions.

Retention Rate

Retention rate is a product metric that measures the share of software users who continue using an application over a defined period.

Recursive Reward Modeling

Recursive reward modeling is an AI training approach that repeatedly improves how software systems evaluate and reward model outputs.

Representation Engineering

Representation engineering is an AI interpretability technique that reads and steers high-level patterns inside neural network activations.

Reward Hacking

Reward hacking is an AI alignment failure in which a system maximizes a scoring rule without completing the intended software task.

Regression to the Mean

Regression to the mean is a statistical effect that makes unusually high or low software measurements tend to be followed by more typical results.

RFM Analysis

RFM analysis is a customer-data segmentation method that scores users by recency, frequency, and monetary value to support software-driven targeting.

Red Team Specialist

A red team specialist is a cybersecurity professional who simulates realistic attacks to test software systems, cloud environments and security controls.

Research Engineer

A research engineer is a software professional who turns experimental ideas into reproducible, testable, and scalable technical systems.

Research Scientist

A research scientist is a technology research professional who designs, tests, and communicates new methods, models, or software capabilities.

Reward Shaping

Reward shaping is a reinforcement learning technique that adds feedback signals to guide an AI agent toward a software task goal.

Reward Function

A reward function is a machine learning function that scores an AI agent's actions to guide it toward a software-defined goal.

ReLU

ReLU is an activation function that turns negative neural-network values into zero while retaining positive values.

Regularization

Regularization is a machine learning technique that limits model complexity to improve predictions on new data.

Random Forest

Random forest is a supervised machine learning algorithm that combines many decision trees to produce more stable predictions.

ROC curve

ROC curve is a model evaluation chart that shows how well a binary classifier separates positive and negative outcomes at different thresholds.

Random Search

Random search is a gradient-free optimization method that tests randomly sampled parameter combinations to find a high-performing software or machine learning configuration.

Regression Model

A regression model is a statistical and machine learning model that estimates a numeric outcome from one or more input variables.

Random Seed

A random seed is a value that initializes a pseudorandom number generator so it produces a repeatable sequence of outputs.

Reward Model

A reward model is a machine learning model that scores AI outputs according to learned human preferences or defined quality rules.

Reinforcement Learning from AI Feedback

Reinforcement learning from AI feedback is a machine learning approach that uses AI-generated evaluations to improve an AI model’s outputs.

Reinforcement Learning from Human Feedback

Reinforcement learning from human feedback is a machine learning method that uses human preferences to guide an AI model's behavior.

Reciprocal Rank Fusion

Reciprocal rank fusion is a search-ranking method that combines several result lists into one ranking using each item’s position.

Retrieval Latency

Retrieval latency is a software performance metric that measures how long a system takes to find and return requested data.

RAG Evaluation

RAG evaluation is a software testing process that measures whether a retrieval-augmented generation system finds useful sources and produces grounded answers.

Recall at K

Recall at K is a retrieval evaluation metric that measures how many relevant results a system finds within its first K results.

Recursive Chunking

Recursive chunking is a text-splitting method that divides large documents at natural boundaries until each piece fits a chosen size limit.

Retriever

A retriever is a software component that finds and returns relevant information from a data source in response to a query.

ReAct prompting

ReAct prompting is an AI prompting method that alternates reasoning, tool actions, and observations to complete multi-step software tasks.

Remote Procedure Call

Remote procedure call is a software communication method that lets one program run a function on another system over a network.

Reflection Loop

Reflection loop is an AI agent pattern that generates an output, evaluates it against evidence or tests, and revises it iteratively.

ReAct agent

A ReAct agent is an AI agent that alternates between reasoning, using tools, and evaluating results to complete a task.

Review Bottleneck

Review bottleneck is a software delivery constraint that delays changes when pull requests arrive faster than qualified reviewers can assess them.

REPL loop

A REPL loop is an interactive programming environment that reads code, evaluates it, prints the result, and waits for the next command.

Role Prompting

Role prompting is a prompt engineering technique that assigns an AI model a defined professional role to guide its responses.

Reranking

Reranking is a second-stage search technique that reorders retrieved results using a more detailed relevance assessment.

Request Body

A request body is an HTTP message component that carries data from a client application to a server.

Refresh Token

A refresh token is a long-lived credential that lets an application obtain new access tokens without asking the user to sign in again.

Runaway Agent

A runaway agent is an autonomous software agent that keeps taking actions or consuming resources without reaching its intended stop condition.

REST API

REST API is an application programming interface that lets separate software systems exchange resource data using standard web requests.

Retrieval-Augmented Generation

Retrieval-augmented generation is an AI method that retrieves relevant external information before an LLM writes an answer.

Reasoning Model

A reasoning model is an AI model that works through multi-step problems before producing an answer or taking an action.

Robotic Process Automation

Robotic process automation is software technology that uses bots to perform repetitive, rule-based digital tasks across business applications.

Rate Limit

A rate limit is a cap on how many requests an app or service will accept in a set time, used to prevent overload and abuse.

Rollup Field

A rollup field is a field that calculates a single value, like a sum or count, from a set of linked records in another table.

Retry Logic

Retry logic is code that automatically re-attempts a failed request, so a temporary glitch doesn't break the whole action.

Rollback

A rollback is the act of returning an app or system to an earlier working state, undoing changes that caused a problem.

S

Staging Environment

A staging environment is a pre-production software environment that closely mirrors the live application for final release validation.

Spot Instances

Spot instances are discounted cloud virtual machines that use spare provider capacity and may be interrupted when that capacity is needed.

Service Level Indicator

Service level indicator is a software reliability metric that measures whether a service delivers an acceptable user experience.

Service Level Objective

A service level objective is a measurable reliability target that defines how well a software service should perform over time.

Service Level Agreement

A service level agreement is a software service contract that defines expected performance, measurement methods, responsibilities, and remedies for missed targets.

Service Mesh

Service mesh is a software infrastructure layer that manages secure, reliable communication between services in a distributed application.

Serverless Computing

Serverless computing is a cloud computing model that runs application code on demand while the provider manages the underlying servers and scaling.

Screen Flow

Screen flow is a user interface workflow that guides a person through connected application screens to complete a task.

Session Tracking

Session tracking is a software technique that links related requests so an application can preserve a user's temporary state.

Semantic Versioning

Semantic versioning is a software release convention that communicates whether an update is breaking, additive, or corrective.

Splash Screen

Splash screen is a temporary startup screen that appears while a software application initializes before its usable interface opens.

SwiftUI

SwiftUI is a declarative user interface framework that lets developers build Apple-platform app screens with Swift code.

Swift

Swift is a programming language that helps developers build fast, safe applications for Apple platforms, servers, and other systems.

Static Hosting

Static hosting is a web hosting approach that delivers pre-built website files directly to visitors without running server-side code for each request.

Subdomain

A subdomain is a DNS-based hostname that creates a distinct address and delivery route under a parent domain.

SSL certificate

An SSL certificate is a digital certificate that verifies a website’s identity and enables encrypted HTTPS connections.

Same-Origin Policy

Same-origin policy is a browser security rule that limits how web pages and scripts interact with data from other origins.

Server-Sent Events

Server-sent events are a web technology that lets a server continuously push one-way updates to a browser over HTTP.

sessionStorage

sessionStorage is a browser Web Storage API that temporarily saves key-value data for one tab and its page session.

State Management

State management is a software engineering practice that stores, updates, and synchronizes changing application data across an interface.

Sass

Sass is a CSS preprocessor that adds reusable, organized authoring features and compiles them into standard browser-ready CSS.

Svelte

Svelte is a front-end framework and compiler that turns component code into efficient browser JavaScript at build time.

Shadow DOM

Shadow DOM is a web platform feature that isolates a component's internal HTML and CSS from the rest of a page.

Service worker

A service worker is a browser-based JavaScript worker that controls network requests to support offline access, caching, and background tasks.

Static Site Generation

Static site generation is a web development approach that builds ready-to-serve HTML pages before visitors request them.

Server-Side Rendering

Server-side rendering is a web development technique that generates a page's HTML on a server before sending it to a browser.

Single-Page Application

A single-page application is a web application that updates views in the browser without loading a new HTML page for each navigation.

Semantic HTML

Semantic HTML is a web development practice that uses meaningful elements to describe content structure and function.

Streaming Data

Streaming data is a continuous flow of software-generated events that systems process as each record arrives.

Shadow Deployment

Shadow deployment is a software release strategy that runs a new version alongside production using mirrored live traffic without affecting users.

Sentence Segmentation

Sentence segmentation is a language-processing and early-literacy practice that divides speech or text into distinct sentence or word units.

Slot Filling

Slot filling is a conversational AI technique that extracts and collects the details needed to complete a user’s request.

Semantic Role Labeling

Semantic role labeling is a natural language processing method that identifies who did what, to whom, when, where, and why in a sentence.

Stop Words

Stop words are common words that text-processing systems may ignore to focus on terms that better identify a document's subject.

Stemming

Stemming is a natural language processing technique that reduces related word forms to a shared shortened stem for easier text analysis.

Spell Checking

Spell checking is a software feature that identifies words that may be misspelled and suggests possible corrections.

Sentiment Analysis

Sentiment analysis is a natural language processing technique that identifies and classifies opinions and emotional tone in text.

Siamese Network

A Siamese network is a neural network architecture that compares inputs by mapping them into a shared similarity space.

System Design Interview

System design interview is a technical interview that evaluates how a candidate designs and explains a scalable software system.

Staff Engineer

A staff engineer is a senior individual contributor who guides software architecture and solves high-impact technical problems across teams.

Software Engineer

A software engineer is a technology professional who designs, builds, tests, and maintains reliable software systems.

Site Reliability Engineer

A site reliability engineer is a software engineer who uses automation, monitoring, and reliability targets to keep online systems dependable.

Solutions Engineer

A solutions engineer is a customer-facing technical professional who maps software products to a prospect's technical needs and demonstrates a workable solution.

Solutions Architect

A solutions architect is a technology professional who designs software systems that meet defined user, technical, security, and operational needs.

Screen Understanding

Screen understanding is a software capability that interprets a digital interface’s visible elements, structure, and likely actions.

Semantic Memory

Semantic memory is an AI knowledge layer that stores reusable facts, concepts, meanings, and relationships independently of individual interactions.

Sycophancy

Sycophancy is an AI model behavior that favors agreement and praise over accurate, evidence-based responses.

State Space Model

A state space model is a mathematical model that represents how a system changes over time through internal states, inputs, and observable outputs.

Sparse Attention

Sparse attention is a transformer attention method that limits each token to a selected subset of other tokens, reducing long-context computing costs.

Self-Attention

Self-attention is a neural network mechanism that lets each item in a sequence weigh the relevance of other items to understand context.

Simulation Testing

Simulation testing is a software and systems testing method that uses a controlled virtual model to evaluate behavior before real-world release.

Synthetic Users

Synthetic users are AI-generated user models that simulate how defined groups of people may think, decide, and respond to products or messages.

Smart Home Automation

Smart home automation is a residential technology system that monitors, controls, and coordinates household devices automatically or remotely.

Spatial Computing

Spatial computing is a computing approach that lets digital systems understand physical space and place interactive digital content within it.

Seasonality

Seasonality is a recurring pattern in data or activity that follows predictable calendar-based intervals.

Sample Size

Sample size is a statistical measure that counts the observations selected from a larger population for analysis.

Sandbagging

Sandbagging is a strategic practice that hides or understates true ability, performance, expectations, or intentions to gain an advantage.

Superposition

Superposition is a principle that combines independent states, waves, or responses to describe a single overall result.

Standardization

Standardization is a business and technical practice that creates consistent rules, specifications, and methods for repeatable results.

Saddle Point

A saddle point is a critical point on a surface where a function rises in some directions and falls in others.

Scrum

Scrum is an agile framework that helps small, cross-functional teams solve complex problems and deliver useful product improvements in short cycles.

Sales-Led Growth

Sales-led growth is a go-to-market strategy that uses salespeople and relationship-based processes to win, expand, and retain customers.

Software as a Service

Software as a Service is a cloud computing model that delivers ready-to-use software over the internet for a recurring fee.

Scalable Oversight

Scalable oversight is an AI safety approach that helps people reliably supervise systems whose work may exceed human ability to check directly.

Statistical Significance

Statistical significance is a statistical measure that indicates whether an observed result is unlikely to arise from random variation alone.

Search Algorithm

A search algorithm is a step-by-step method that finds a requested item, path, or best solution within data or a problem space.

Symbolic AI

Symbolic AI is an artificial intelligence approach that uses explicit symbols, rules, and logic to reason through problems.

Swarm Intelligence

Swarm intelligence is a decentralized AI approach that uses simple agents and local interactions to solve complex group problems.

Sequence Modeling

Sequence modeling is a machine learning approach that learns patterns in ordered data so it can interpret, predict, or generate what comes next.

Self-Supervised Learning

Self-supervised learning is a machine learning approach that creates training signals from raw data to learn useful patterns without manual labels.

Semi-Supervised Learning

Semi-supervised learning is a machine learning approach that trains models with a small labeled dataset and a larger unlabeled dataset.

Supervised Learning

Supervised learning is a machine learning method that learns from labeled examples to predict a known outcome for new data.

Software 3.0

Software 3.0 is an AI-centered programming paradigm that uses natural-language instructions, context, and tools to produce software outcomes.

Software 2.0

Software 2.0 is a machine-learning approach that creates program behavior by training neural networks on data rather than writing every rule by hand.

Small Language Model

A small language model is a compact AI model that understands and generates text while using less computing power than larger language models.

Shadow IT

Shadow IT is unapproved technology used for work that operates outside an organization's formal IT and security oversight.

Spring Boot

Spring Boot is a Java framework that simplifies building, running, and operating production-ready Spring applications.

Strong Consistency

Strong consistency is a data consistency model that makes every read reflect the latest committed write across a distributed system.

Sharding

Sharding is a database scaling technique that splits data across independent servers to handle more storage and traffic.

SQL

SQL is a database query language that defines, retrieves, changes, and protects structured data stored in relational databases.

Supabase

Supabase is an open-source backend-as-a-service platform that provides Postgres databases, authentication, storage, APIs, and realtime features.

SQLite

SQLite is an embedded relational database engine that stores structured application data in a single local file.

Server-Side Caching

Server-side caching is a software performance technique that stores reusable responses or data near an application server for faster repeat requests.

Session Management

Session management is a web application process that recognizes users across requests and controls their authenticated access over time.

Soft Delete

Soft delete is a database pattern that marks data as deleted while retaining it for recovery, auditing, or scheduled removal.

Serverless Database

A serverless database is a cloud database service that automatically manages infrastructure, scaling, and routine maintenance as demand changes.

SQL JOIN

SQL JOIN is a database query operation that combines related rows from two or more tables using matching column values.

Sample Efficiency

Sample efficiency is a machine learning property that measures how well a model learns from limited data or interactions.

Self-Instruct

Self-Instruct is an AI training framework that creates and filters synthetic instructions to improve a language model's ability to follow requests.

Self-training

Self-training is a semi-supervised machine learning method that expands a training set with a model’s high-confidence predictions.

Synthetic Data

Synthetic data is artificially generated data that imitates the structure and patterns of real software datasets without using the original records.

Supervised Fine-Tuning

Supervised fine-tuning is a machine learning method that adapts a pre-trained model using curated input-output examples.

Secondary Sale

Secondary sale is a software licensing transaction that transfers an existing license or digital entitlement from its current holder to another user.

Startup Acquisition

Startup acquisition is a technology business transaction that transfers ownership of a software startup to a buyer seeking its product, code, users, or expertise.

SAFE Note

A SAFE note is a startup financing agreement that lets a software company raise funds now for future equity.

Startup Valuation

Startup valuation is a financial estimate that sets the implied worth of a technology startup for an investment or equity transaction.

Series B

Series B is a venture capital funding round that helps a software startup scale a proven product, technical platform, and customer base.

Series A

Series A is a startup funding round that helps a software company scale a validated product, technical infrastructure, and customer growth.

Seed Funding

Seed funding is early-stage capital that helps a software startup build, test, and validate an initial product.

Sales Pipeline

Sales pipeline is a software-based workflow that tracks revenue opportunities through defined stages from captured lead to closed outcome.

T

Trunk-Based Development

Trunk-based development is a software development workflow that keeps code integrated in one shared main branch through small, frequent changes.

Terraform

Terraform is an infrastructure as code tool that defines, provisions, and updates cloud and on-premises resources through versioned configuration files.

Tauri

Tauri is an open-source application framework that builds desktop and mobile apps with web frontends and a Rust-based native layer.

Thumb Zone

Thumb zone is a mobile UX design concept that maps the screen areas a user can comfortably reach with one thumb.

TestFlight

TestFlight is an Apple beta testing platform that lets developers distribute pre-release app builds and collect tester feedback before App Store release.

Theming

Theming is a software design practice that applies shared visual rules to change an application's appearance without changing its core behavior.

Time to First Byte

Time to First Byte is a web performance metric that measures how long a browser waits to receive a server response.

TLS

TLS is a cryptographic protocol that encrypts and authenticates data sent between software clients and servers over untrusted networks.

Tree Shaking

Tree shaking is a build-time optimization that removes unused code from a software application's production bundle.

Transpiler

A transpiler is a source-to-source compiler that converts code from one programming language or syntax into another readable form.

Tailwind CSS

Tailwind CSS is a utility-first CSS framework that lets developers build custom, responsive web interfaces by composing small styling classes in markup.

TypeScript

TypeScript is a programming language that adds type checking to JavaScript and compiles into JavaScript for browsers and servers.

Tenant Isolation

Tenant isolation is a software security practice that keeps each customer’s data, access, and workloads separated in a shared application.

Training Pipeline

Training pipeline is a machine learning workflow that prepares data, trains and evaluates a model, then records an approved version for use.

TruthfulQA

TruthfulQA is an AI benchmark that tests whether language models give truthful answers instead of repeating common falsehoods.

Text Normalization

Text normalization is a data-preparation process that converts varied text into a consistent form for reliable search, analysis, and language processing.

Text Clustering

Text clustering is an unsupervised machine learning technique that groups similar documents, sentences, or phrases without predefined labels.

Topic Modeling

Topic modeling is an unsupervised machine learning method that finds recurring themes in a large collection of text.

Treebank

A treebank is a linguistically annotated text corpus that records how words and phrases relate within sentences.

Text Corpus

A text corpus is a structured collection of written language that supports systematic analysis, search, and language technology training.

TF-IDF

TF-IDF is a text-analysis method that scores words by how important they are within one document relative to a collection.

Text Simplification

Text simplification is a language process that makes written information easier to read while preserving its essential meaning.

Text Generation

Text generation is an AI capability that produces written language from prompts, data, or other input.

Text Summarization

Text summarization is a natural language processing technique that condenses source text while preserving its central ideas and relevant facts.

Text Classification

Text classification is a natural language processing task that assigns predefined labels to written content based on its meaning.

Total Compensation

Total compensation is a technology-sector pay framework that measures the full value of a software professional's cash pay, equity, benefits, and role-related rewards.

Take-Home Assignment

Take-home assignment is a software engineering interview assessment that lets candidates complete a defined technical task independently.

Technical Interview

Technical interview is a software engineering assessment that evaluates how a candidate solves technical problems, explains decisions, and applies relevant skills.

Tech Lead

A tech lead is a senior software engineer who guides a team's technical decisions, quality standards, and delivery work.

Technical Account Manager

A technical account manager is a customer-facing technology professional who helps clients implement, operate, and improve software products.

Tool Misuse

Tool misuse is an AI agent security risk that causes authorized software tools to perform unintended, unsafe, or harmful actions.

Termination Condition

Termination condition is a software rule that determines when a loop, workflow, algorithm, or agent process must stop.

Tool Discovery

Tool discovery is a software capability that finds and selects the right API, service, function, or agent tool for a task.

Tool Schema

Tool schema is a machine-readable software contract that tells an AI agent how to call a specific tool safely and correctly.

Token Economy

Token economy is a digital incentive and value system that defines how a platform creates, distributes, uses, and governs its tokens.

Transformer Architecture

Transformer architecture is a neural network design that uses attention to understand relationships across text, images, code, and other sequences.

TinyML

TinyML is a machine learning approach that runs compact AI models directly on low-power devices such as microcontrollers.

Time Series Analysis

Time series analysis is a statistical method that examines time-ordered data to identify patterns, explain change, and forecast likely future values.

Turing test

Turing test is an AI evaluation method that tests whether a machine can converse convincingly enough to be mistaken for a human.

Technological Singularity

Technological singularity is a hypothetical future point where AI-driven progress becomes faster than humans can predict or control.

Transfer Learning

Transfer learning is a machine learning technique that reuses knowledge from a trained model to solve a related new task.

Thread Pool

A thread pool is a concurrency mechanism that reuses a limited set of worker threads to run queued software tasks.

Time-Series Database

A time-series database is a specialized database that stores, indexes, and analyzes data points recorded over time.

Training FLOPs

Training FLOPs is a compute estimate that measures the floating-point operations needed to train an AI model.

Training Stability

Training stability is a machine learning property that keeps model learning reliable as training data, settings, and iterations change.

Training Loss

Training loss is a machine learning metric that measures how far a model's predictions are from correct training data labels.

Training Tokens

Training tokens are tokenized units of text, code, or other data that an AI model processes to learn patterns.

Tensor Parallelism

Tensor parallelism is a model-parallel computing technique that splits neural network tensors and layer calculations across multiple processors.

Training Run

Training run is a machine learning execution that processes data to fit or update a model and records its resulting artifacts and metrics.

Term Sheet

A term sheet is a preliminary technology agreement that summarizes the key product, technical, commercial, and legal terms before a final contract.

Total Addressable Market

Total addressable market is a market-sizing metric that estimates the maximum software revenue opportunity for a defined product category.

Time to Value

Time to value is a software product metric that measures how long users take to achieve a meaningful result after starting to use an application.

Tripwire Capability

Tripwire capability is a security monitoring function that detects defined changes or events in software systems and triggers an alert or response.

t-test

A t-test is a statistical hypothesis test that helps software teams determine whether a difference in numeric results is likely meaningful.

Temporal Difference Learning

Temporal difference learning is a reinforcement learning method that improves predictions from each new software interaction and its next estimated outcome.

tanh

tanh is a hyperbolic activation function that maps any numeric input to an output strictly between -1 and 1.

t-SNE

t-SNE is a dimensionality reduction technique that maps complex, high-dimensional data into a visual two- or three-dimensional layout.

Test Set

A test set is a defined collection of test data or test cases used to evaluate whether software or an AI model works as expected.

Training Data

Training data is a machine learning dataset that teaches an AI model to recognize patterns, predict outcomes, or generate outputs.

Tokens Per Second

Tokens per second is an AI inference metric that measures how many text tokens a language model generates each second.

Top-K Sampling

Top-K sampling is a language model decoding method that limits each next-token choice to a fixed number of likely options.

Top-p Sampling

Top-p sampling is an AI decoding method that selects the next token from a flexible pool of likely options.

Triplet Loss

Triplet loss is a machine learning loss function that trains embeddings to rank matching items closer than nonmatching items.

Test-Time Compute

Test-time compute is an AI inference technique that uses extra processing after training to improve an answer before it is returned.

Thinking Tokens

Thinking tokens are internal AI model tokens that support multi-step reasoning before a model produces its visible answer.

Tree of Thoughts

Tree of Thoughts is an AI reasoning framework that explores, evaluates, and revises multiple solution paths for complex problems.

Tool Registry

A tool registry is a software catalog that records, describes, secures, and helps applications or AI agents find callable tools.

Time to First Token

Time to first token is an LLM latency metric that measures how long an application waits before streaming its first response token.

Triage Agent

A triage agent is an AI software agent that classifies, prioritizes, investigates, and routes incoming technical requests or alerts.

Task decomposition

Task decomposition is a software engineering practice that breaks a complex technical goal into smaller, testable, dependent or independent subtasks.

Template Marketplace

A template marketplace is a software platform that lists, distributes, and sometimes sells reusable digital templates for building websites, apps, and workflows.

Task Automation

Task automation is a software capability that completes defined, repeatable digital actions with minimal human input.

Third-Party Integration

Third-party integration is a software connection that lets an application exchange data or trigger actions with an external service.

Token

A token is a small chunk of text an AI model reads and writes in, and it's the unit an agent's memory, or context window, is measured in.

U

URL Redirect

URL redirect is a web routing mechanism that automatically sends a browser or search engine from one address to another.

URL slug

A URL slug is a readable URL component that identifies a specific page or resource within a website.

Utterance

Utterance is a unit of spoken or written expression that conveys meaning within a particular communicative context.

U-Net

U-Net is a neural network architecture that labels every relevant pixel in an image to create precise segmentation maps.

Upskilling

Upskilling is a technology skills-development process that expands a person's ability to build, maintain, test, or use software effectively.

User Prompt

A user prompt is an instruction or input that tells an AI system what task to perform or response to generate.

Universal Jailbreak

Universal jailbreak is an AI security attack that uses one reusable input pattern to bypass safeguards across many restricted requests or models.

Unit Economics

Unit economics is a financial analysis method that measures the revenue, direct costs, and profit associated with one business unit.

Unsupervised Learning

Unsupervised learning is a machine learning approach that finds patterns, groups, and relationships in unlabeled data.

Upsell

Upsell is a software product strategy that guides users toward a higher-value plan, feature tier, or paid add-on that better fits their needs.

User Story

A user story is a lightweight software requirement that describes a user goal and the value a feature should deliver.

User Onboarding

User onboarding is a software experience that guides new users to understand an application and achieve an early valuable outcome.

Usage-Based Pricing

Usage-based pricing is a software pricing model that charges customers according to measurable product consumption.

Undersampling

Undersampling is a data or signal-processing technique that reduces samples to manage imbalance, processing cost, or sampling requirements.

UMAP

UMAP is a dimensionality reduction technique that maps complex data into fewer dimensions while preserving meaningful relationships.

Unlabeled Data

Unlabeled data is a machine learning dataset that contains inputs without the correct categories, outcomes, or target answers attached.

Underfitting

Underfitting is a machine learning problem that occurs when a model is too simple to learn meaningful patterns from data.

Upsert

Upsert is a database operation that creates a record when it is missing or updates it when a matching record already exists.

Unattended Automation

Unattended automation is a software automation approach that runs scheduled or event-triggered workflows without a person operating each step.

Universal Key

Universal Key is one Emergent credential that connects your app to multiple LLMs like GPT, Claude, and Gemini, billed in Emergent credits.

V

Version Control

Version control is a software engineering system that records file changes so teams can collaborate, review work, and restore earlier code safely.

Virtual Machine

A virtual machine is a software-defined computer that runs its own operating system and applications on shared physical hardware.

Virtual Private Cloud

A virtual private cloud is a logically isolated cloud network that lets teams control how application resources communicate.

Vite

Vite is a frontend build tool that runs a fast development server and creates optimized production files for modern web applications.

Virtual DOM

Virtual DOM is a software rendering technique that represents a user interface in memory before efficiently updating the browser DOM.

Vue.js

Vue.js is a progressive JavaScript framework that helps developers build interactive web interfaces from reusable components.

Viewport

Viewport is a browser viewing area that determines which part of a web page is visible and how responsive layouts adapt.

Variational Autoencoder

A variational autoencoder is a generative machine learning model that learns a probabilistic, compressed representation of data to reconstruct and create similar examples.

Vibe Debugging

Vibe debugging is an AI-assisted software debugging practice that uses natural-language prompts, runtime evidence, and testing to diagnose and fix application defects.

Vision-Language Model

A vision-language model is an AI model that connects images or video with text to interpret, answer questions about, and generate language from visual content.

Vertical AI

Vertical AI is a specialized artificial intelligence system that supports the data, rules, workflows, and decisions of a specific industry or function.

Virtual Reality

Virtual reality is an immersive computing technology that places users inside interactive, computer-generated three-dimensional environments.

Venture Capital

Venture capital is a form of private equity financing that funds high-growth startups in exchange for an ownership stake.

Vertical Scaling

Vertical scaling is a software infrastructure approach that increases the computing capacity of one server or application instance.

Validation Loss

Validation loss is a machine learning metric that measures a model's prediction error on separate data not used to train it.

Vesting Cliff

Vesting cliff is a software startup equity milestone that delays ownership until a defined date for founders and technical hires.

Vesting Schedule

A vesting schedule is a software entitlement rule that gradually grants a user ownership, access, or token rights over time or after milestones.

Viral Coefficient

Viral coefficient is a product growth metric that estimates how many new users each existing user generates through a software product.

Viral Loop

Viral loop is a software growth mechanism that turns a user's product action into an invitation or artifact that brings in new users.

Value Proposition

Value proposition is a software product positioning statement that explains who an application helps, what problem it solves, and why it is preferable.

Value Function

Value function is a machine learning function that estimates the expected future reward of a software agent's state or action.

Vanishing Gradient Problem

Vanishing gradient problem is a deep-learning training failure that makes early neural-network layers receive updates too small to learn.

Validation Set

Validation set is a machine learning data subset that guides model selection and tuning during development.

Vocabulary Size

Vocabulary size is a natural language processing parameter that defines how many distinct tokens a software system can recognize or generate.

Vector Namespace

Vector namespace is a logical partition that separates embeddings within a vector database for targeted storage, retrieval, and management.

Vector Index

Vector index is a software data structure that organizes embeddings so applications can quickly find similar items.

Voice Agent

A voice agent is an AI software agent that understands spoken requests, responds aloud, and can use connected tools to complete tasks.

Vendor Lock-In

Vendor lock-in is a technology dependency that makes it difficult or costly to move software, data, or workflows to another provider.

Vibe Coding

Vibe coding is an AI-assisted software development approach that turns plain-language requests into working application code.

Vector Database

A vector database is a specialized database that stores embeddings and finds information by semantic similarity rather than exact matches.

Vector Search

Vector search is a retrieval method that finds meaningfully similar items by comparing their numerical embeddings.

Visual Edits

Visual Edits lets you click any element in your Emergent app preview, say the change, and stack edits to apply to the real code in one batch.

Voice Mode

Voice Mode lets you build apps on Emergent by speaking, turning your spoken description into working code and UI in real time.

W

Web Manifest

A web manifest is a JSON file that tells browsers how a web application should appear and launch when installed.

WebP

WebP is a raster image format that compresses web images efficiently while supporting transparency, animation, and lossy or lossless encoding.

WordPress

WordPress is an open-source content management system that lets people create, publish, and manage websites without building every page from scratch.

Web Server

A web server is software that receives web requests and returns website files, application data, or API responses over HTTP or HTTPS.

Web Hosting

Web hosting is a technology service that stores a website's files and delivers them to visitors over the internet.

WebSocket

WebSocket is a communication protocol that keeps a two-way connection open between a client application and server.

Webpack

Webpack is a module bundler that analyzes application files and packages them into browser-ready assets.

Web Components

Web Components are browser-native UI building blocks that package reusable custom HTML, behavior, and styling for use across web applications.

WCAG

WCAG is a web accessibility standard that defines testable requirements for making digital content usable by people with disabilities.

Web Accessibility

Web accessibility is a software design and development practice that makes websites and web apps usable with diverse abilities and assistive technologies.

Workflow Orchestration

Workflow orchestration is a software engineering practice that coordinates dependent tasks, services, and data into reliable end-to-end processes.

Win Rate

Win rate is a performance metric that shows the share of attempts, opportunities, or contests that end in a successful outcome.

WordNet

WordNet is a lexical database that groups English words by meaning and connects those meanings through defined semantic relationships.

Weight Sharing

Weight sharing is a neural network technique that reuses the same learned parameters across multiple parts of a model.

Wearable AI

Wearable AI is a category of body-worn technology that uses artificial intelligence to interpret data and provide timely assistance.

World Model

A world model is an AI system that builds an internal representation of an environment to predict how it changes and responds to actions.

Worker Process

Worker process is a software process that runs assigned tasks separately from an application's main request-handling or user-interface process.

Wide-Column Store

A wide-column store is a NoSQL database that organizes flexible, sparsely populated data into column families for scalable, key-based access.

Weight Decay

Weight decay is a machine learning regularization technique that discourages overly large model parameters during training.

Weekly Active Users

Weekly active users is a software product metric that counts unique users who complete a defined meaningful action within seven days.

Waitlist Launch

Waitlist launch is a software product release strategy that collects early signups before granting access to an application.

Weak-to-Strong Generalization

Weak-to-strong generalization is an AI training approach that helps a more capable model learn useful behavior from weaker, imperfect supervision.

Weaviate

Weaviate is a vector database that stores data and embeddings so software can find related content by meaning.

Weight Initialization

Weight initialization is a machine learning technique that sets a neural network's starting trainable values before learning begins.

WordPiece

WordPiece is a subword tokenization algorithm that converts text into vocabulary tokens a language model can process.

Word2Vec

Word2Vec is a word embedding technique that learns numerical representations of words from the contexts in which they appear.

Word Embedding

Word embedding is a machine learning representation that converts words into numeric vectors that preserve useful relationships in text.

Workflow Monitoring

Workflow monitoring is a software practice that tracks automated workflow execution, health, and results so teams can detect and resolve issues.

Webhook Signature

A webhook signature is a cryptographic check that verifies a webhook request came from its expected sender and was not changed.

WYSIWYG Editor

A WYSIWYG editor is a visual software editor that lets users format digital content while seeing an on-screen approximation of the final result.

Web Scraping

Web scraping is a software technique that automatically collects website data and converts it into structured, reusable information.

White-Label App

A white-label app is a software application that one provider builds for others to brand and offer as their own.

Workflow Automation Agent

A workflow automation agent is a software agent that interprets inputs, chooses permitted actions, and completes multi-step tasks across connected applications.

Workflow

Workflow is a repeatable process that organizes people, tasks, information, and decisions to complete a specific outcome.

Webhook

A webhook is an automatic message one app sends another the moment an event happens, so the second app can react without checking for updates.

Start Building
on Emergent today
Start Building

Build production-ready apps through conversation. Chat with AI agents that design, code, and deploy your application from start to finish.

SOC 2 certificationISO 27001 certification
Product
PricingAI Website BuilderAI Landing Page BuilderAI Form BuilderAI Survey and Quiz BuilderAI App BuilderAI Web App BuilderAI Dashboard BuilderAI CRM BuilderAI SaaS BuilderAI Schedule BuilderAI Automation BuilderAI Agent BuilderAI Chatbot Builder
Resources
DocsBlogTutorialsCase StudiesLearnVideosNewsIntegrationsGlossaryCommunityAI ContestsSitemap
Solutions
EnterpriseIT AgenciesSMB OwnersProduct ManagersOperations Team
Company
AffiliatesCareersInfoPrivacy PolicyCookie PolicyTerms of ServiceDPASub ProcessorsIP Addresses
Copyright
Copyright icon
Emergentlabs 2026
Designed and built by
the awesome people of Emergent 🩵