Getting Started with MoltBot on Emergent

This tutorial walks through the complete beginner journey of setting up MoltBot on Emergent and then extending it to real-world communication channels like Telegram and WhatsApp.

By the end, you will not only have MoltBot running, but also be able to interact with it outside the Emergent interface and understand how to keep it running autonomously.

Introduction

MoltBot is a preconfigured autonomous agent chip on Emergent that can be launched in minutes. Traditionally, setting up an autonomous agent required installing dependencies, configuring API keys, provisioning servers, and accepting security risks on a personal machine or paid VPS.

Emergent removes these barriers by handling the entire setup automatically in the cloud. A single action provisions the agent runtime, connects language models, and prepares the environment for multi-platform interaction.

This tutorial covers the initial setup and the most common next steps users take to get real value from MoltBot. If you want to see the full walkthrough of using Moltbot on Emergent, do check out the video down below!

TLDR: Essentials to setup Moltbot with Emergent

Here's the key steps and considerations you need to follow to start exploring straight away:

• If you haven't yet, you should definitely set up your Universal Key on Emergent to automatically refill the balance once it drops below a certain number of credits (we recommend 5). Here's where to find those settings:

• If you have an existing Claude subscription, you would see more economical token rates if you use your own API key. Simply select the Anthropic option from the initial dropdown where you select which key to use with Moltbot, and enter your own key.
  
  We've outlined how to get the same down below!

  
  

• If you want your Moldbot to be persistent, and stay online 24x7, do remember to publish the bot with the "Publish" button on the top right of your Emergent chat. Again, we cover this further ahead too.

Learning Outcomes

By completing this tutorial, you will be able to:

• Launch MoltBot with no manual configuration

• Understand how Emergent provisions the agent environment

• Connect MoltBot to Telegram

• Connect MoltBot to WhatsApp

• Interact with the agent across platforms

• Publish the agent so it runs continuously

• Use the agent for basic automation and scheduled tasks

Step 0: Prerequisites

Before starting, ensure the following:

• A modern web browser such as Chrome

• A Google account for authentication

• A Telegram account if you plan to connect Telegram

• A WhatsApp account if you plan to connect WhatsApp

No API keys are required to get started.

Step 1: Open the Emergent Home Page

Navigate to:

https://emergent.sh

This is the main entry point for launching agent chips.

Step 2: Select the MoltBot Chip

Locate the MoltBot chip on the home screen and select it.

This chip represents a predefined autonomous agent template managed by Emergent.

Step 3: Initiate the Setup

With the MoltBot chip selected, press Enter or trigger the primary action.

This single action starts the entire setup flow. No configuration screens or forms appear.

Step 4: Autonomous Background Setup

Emergent now provisions everything automatically in the background:

• Agent runtime on a cloud virtual machine

• MoltBot dependencies and tools

• Default recommended LLM configuration

• Secure, isolated execution environment

This process usually completes in about five minutes, though it may take between two and eight minutes.

No interaction is required during this phase.

Step 5: LLM Key Handling

Emergent supplies its own recommended LLM keys by default. This allows users to start immediately without providing OpenAI or Anthropic credentials.

Users can optionally add their own API keys later, but this is not required to proceed.

Optional: Using Your Own Anthropic API Key via Claude Terminal

Advanced users who have an Anthropic Max plan may prefer to use their own Anthropic API key instead of the default key provided by Emergent.

Anthropic provides a way to generate API tokens directly from the Claude CLI. This process must be done on the user’s local machine and is completely optional.

When This Is Useful

You may want to bring your own Anthropic API key if:

• You already have an Anthropic Max subscription

• You want full control over usage limits and billing

• You are testing specific Claude models or configurations

• You prefer separating personal API usage from Emergent defaults

For most users, the default Emergent-provided key is sufficient and no action is required.

Generating an Anthropic API Key Using Claude CLI (For Power Users with Claude Max)

This process is handled locally on your computer.

Prerequisites

Before starting, ensure that:

• You have an active Anthropic Max account

• You have installed the Claude CLI locally

• You are logged in to your Anthropic account in the terminal

This step does not involve Emergent directly.

Step 1: Open Your Local Terminal

Open a terminal on your local machine. This can be Terminal on macOS, Command Prompt or PowerShell on Windows, or any Linux terminal.

Step 2: Run the Token Setup Command

Run the following command:

claude setup-token

This command initiates the token generation flow for your Anthropic account.

You may be prompted to authenticate or confirm access depending on your setup.

Step 3: Copy the Generated API Key

Once the command completes, the CLI will output an Anthropic API key.

Copy this key securely. Treat it like a password.

Do not commit this key to version control or share it publicly.

Step 4: Add the API Key to MoltBot on Emergent

Return to the MoltBot interface on Emergent.

In the model or API configuration section:

• Select Anthropic as the provider

• Paste the API key generated from the Claude CLI

• Save or confirm the configuration

MoltBot will now use your personal Anthropic key instead of the default Emergent-managed key.

Important Notes and Limitations

• This process must be run locally. It cannot be executed inside the Emergent environment.

• Billing and rate limits are controlled by your Anthropic account.

• If the key is removed or expires, MoltBot will fall back to requiring a valid key.

• Switching keys does not affect agent memory or integrations.

But continuing onwards, for the folks who want to go ahead with the Emergent Key.

Step 6: Setup Completion Confirmation

When setup finishes, the interface displays a success message:

Open Claw started successfully

This confirms that MoltBot is live and the agent environment is ready.

Step 7: Sign In to Continue

Select Continue with Google and sign in.

Any email address can be used. This step creates or links your Emergent account and unlocks the agent workspace.

Step 8: Understanding the Default Agent State

At this point:

• MoltBot is running on a cloud VM

• The agent is isolated from your personal device

• No local computer access is granted

• The agent can access web-based tools and APIs

If the agent is not published, it will stop when the session ends. Publishing is required for long-running tasks and automations.

Step 9: Connecting MoltBot to Telegram

Telegram is the most common first integration.

Create a Telegram Bot

1. Open Telegram on mobile or web

2. Search for @BotFather

3. Send the command /newbot

4. Choose a name for your bot

5. Choose a username ending with bot

6. Copy the API token provided by BotFather

Connect Telegram to MoltBot

1. Return to the Emergent MoltBot interface

2. Paste the Telegram API token into the chat or connection field

3. Find your newly created bot on Telegram

4. Send a message such as hi

5. Receive a pairing code from the bot

6. Enter the pairing code back into Emergent

Once confirmed, the connection is complete.

Test the Integration

Send a message to your Telegram bot, for example:

Send me five random life philosophies

MoltBot should respond directly in Telegram.

Step 10: Connecting MoltBot to WhatsApp

WhatsApp integration allows the agent to communicate via your phone number.

Initiate WhatsApp Connection

1. In the Emergent interface, ask the agent:
   Help me connect my WhatsApp

2. Open the Channels or Integrations section

3. Select WhatsApp

4. Click Show QR

Complete the Connection

1. Open WhatsApp on your phone

2. Scan the QR code

3. If prompted, refresh once

4. Enter your phone number with country code

Once connected, the agent will send a confirmation message to WhatsApp.

Step 11: Cross-Platform Interaction

After connecting Telegram and WhatsApp:

• You can message the agent from any connected platform

• Context is preserved across platforms

• You can instruct the agent to send messages to specific contacts

Example prompt inside Moltbot Control UI (Via Emergent):

Send a summary of today’s AI news to my WhatsApp

Note:

If you face any issues in connecting WhatsApp or Telegram via the agent chat, please navigate to the Channels tab in the Control UI and follow the steps there!

Step 12: Publishing the Agent for Persistence

To enable long-running behavior:

• Publish the agent from the Emergent interface

Publishing keeps the virtual machine running continuously. This is required for:

• Scheduled tasks

• Cron jobs

• Continuous monitoring

• Background automation

Unpublished agents stop when the session ends.

Step 13: Using Scheduled Tasks and Automation

Once published, you can create recurring tasks.

Example prompts:

• Send me the latest AI developments every hour

• Check Twitter for viral AI posts daily and summarize them

• Remind me every morning to practice Spanish

These tasks run automatically without further input.

What MoltBot Can and Cannot Do

What Works Well

• Web-based tools and APIs

• Messaging platforms

• GitHub repositories

• Research and monitoring

• Content generation

• Scheduling and automation

What Does Not Work

• Access to local files or apps on your personal device

• Apple Notes, local folders, or system-level permissions

• Anything requiring physical device access

Ideas to Build With MoltBot

Once MoltBot is set up and connected to your preferred platforms, it becomes a flexible foundation for building personalized, autonomous workflows. Below are example use cases that users can build or adapt based on their own needs.

These are not presets. They are starting points for custom agents.

Daily Morning Briefing

MoltBot can act as a single daily briefing system that replaces dozens of notifications.

Each morning, the agent can:

• Scan emails received in the last 24 hours

• Identify which messages require action or decisions

• Summarize key points instead of forwarding raw emails

• Check real-time traffic conditions for the user’s commute

• Adjust recommended departure time based on traffic

• Set an alarm ten minutes before the user needs to leave

• Curate news updates from topics, industries, or niches chosen by the user

The result is one concise briefing delivered via Telegram, WhatsApp, or another connected channel, instead of fragmented alerts across multiple apps.

This use case works especially well with scheduled tasks and agent publishing enabled.

Memory-Based Personal CRM

MoltBot can function as a long-term memory system for human relationships.

By reading conversations across connected platforms, the agent can:

• Build a structured record of people the user interacts with

• Store personal details such as preferences, context, and past discussions

• Remember notes added explicitly by the user through text or voice

• Retrieve all known information about a person on request

Example prompts include:

• Remember that Alex prefers async communication

• What do I know about Sarah from past conversations

• Summarize my last three interactions with this client

This turns MoltBot into a personal CRM without manual data entry or rigid forms.

Language Learning via Voice Coaching

MoltBot can be used as a continuous language learning companion embedded into daily communication.

Users interact with the agent through voice messages on platforms like Telegram or Slack. MoltBot can:

• Listen to pronunciation and speech patterns

• Compare them against pronunciation datasets

• Identify specific sounds, words, or phrases that need improvement

• Provide targeted feedback instead of generic corrections

• Suggest focused practice based on recurring mistakes

Because this happens inside normal conversation, learning becomes low-friction and continuous rather than session-based. Want to check out how it may work? Here’s our shot at it:

Smart Deal Finder and Negotiator

MoltBot can automate price comparison and negotiation.

In this setup, the agent:

• Searches the web for products or services the user wants to buy

• Identifies relevant sellers and listings

• Contacts sellers via email or website forms

• Requests discounts or negotiates pricing

• Collects and evaluates responses

• Presents the best available option to the user

The user stays out of the negotiation loop until a decision is required, turning deal hunting into a background task.

Trend Monitoring and Auto Execution

MoltBot can continuously monitor trends across the internet.

Users can specify:

• Topics, niches, or industries

• Competitors or brands to watch

• Social platforms, news sources, or RSS feeds

The agent can then:

• Detect emerging trends or viral content

• Summarize what is gaining traction and why

• Suggest app ideas or content angles

• Automatically generate posts in the user’s voice

• Build quick prototypes using Emergent tools

• Publish content automatically in supported workflows

This use case allows users to react to trends in near real time without constant manual monitoring.

How to Approach Building Your Own Use Case

When designing a MoltBot workflow, start by answering three questions:

1. What information should the agent monitor continuously

2. What decisions or actions should happen automatically

3. Where should results be delivered

Because MoltBot runs in a persistent cloud environment, it is best suited for workflows that benefit from memory, scheduling, and cross-platform communication.

These examples are meant to show the range of what is possible rather than prescribe fixed solutions. Each can be combined, simplified, or expanded depending on the user’s goals.

Recap

In this tutorial, you:

• Set up MoltBot in minutes using Emergent

• Avoided manual configuration and API key setup

• Connected the agent to Telegram and WhatsApp

• Learned how to publish the agent for persistence

• Enabled cross-platform communication and automation

MoltBot is now ready to act as a persistent autonomous assistant across messaging platforms.