Hob: Open-Source Local AI Desktop App Launches Today

FrigadeHQ has launched Hob, a free and open-source desktop application that brings AI file processing capabilities to Mac, Windows, and Linux systems while keeping all data strictly local. Unlike cloud-dependent alternatives such as Claude Cowork and ChatGPT desktop apps, Hob runs AI models directly on your computer, ensuring complete privacy and offline functionality.

Release Date and Availability

Officially launched on October 26, 2026, Hob is immediately available for download on all major desktop platforms. The application is distributed under the MIT license, allowing developers to modify, distribute, and deploy the software without restriction. Christian, the creator from FrigadeHQ, announced the project on Hacker News after several weeks of development.

On-Device Processing Architecture

Hob differentiates itself through its completely local processing model. The application currently supports two open-weight AI models: Gemma 4 and GPT-OSS. All conversations, file analyses, and generated content remain on your system, with no data transmitted to external servers during normal operation. This architecture addresses growing privacy concerns around cloud-based AI tools that process sensitive documents and proprietary information.

The only network connections occur when users enable optional web search functionality or integrate MCP connectors. Both features can be disabled in the application settings, allowing for completely offline operation when network isolation is required.

Technical Capabilities and Model Support

The desktop application provides file interaction capabilities similar to commercial alternatives. Users can:

• Upload and analyze documents directly within the chat interface
• Process multiple file types for summarization and questioning
• Run inference tasks without internet connectivity
• Customize model parameters for specific use cases

By leveraging open-weight models, Hob gives users transparency into the underlying AI systems. The Gemma 4 and GPT-OSS models offer competitive performance for common tasks including code analysis, document summarization, and content generation, all while running on consumer hardware.

Privacy and Data Control

The application stores all conversation history and uploaded files locally on the user's machine. This approach eliminates concerns about data retention policies, third-party access, or potential breaches of cloud infrastructure. Organizations handling confidential information can deploy Hob across teams without external data exposure risks.

The MIT licensing also enables enterprises to audit the codebase, verify security implementations, and customize the application for specific compliance requirements. Source code transparency provides assurance that no telemetry or hidden data collection occurs during operation.

What This Means

Hob represents a significant step toward democratizing AI tools while prioritizing user privacy. As concerns mount over how cloud AI services handle sensitive data, locally-run alternatives offer viable paths for individuals and organizations requiring strict data controls. The open-source nature ensures the project can evolve through community contributions, potentially expanding model support and feature sets beyond the initial Gemma 4 and GPT-OSS integration. For developers and privacy-conscious users, Hob provides a foundation for building custom AI workflows without dependency on commercial cloud platforms.