An AI-powered desktop agent for macOS, Fazm automates computer tasks via voice commands. It targets users needing to control browsers, write code, manage documents, operate Google Apps, and learn custom workflows, offering a fully local, open-source solution. The primary benefit is enhanced productivity through voice-driven automation and personalized workflow learning.

How It Works

The core of Fazm is a macOS desktop application built with Swift/SwiftUI. It integrates with AI models, potentially via a TypeScript-based bridge for Claude, to understand and execute user commands. The system operates locally, with billable model keys fetched at runtime and authenticated via a hosted backend. Configuration can be customized by providing a local .env.app file to point to a self-hosted backend.

Quick Start & Requirements

• Primary install/run command: ./run.sh (builds and launches the Swift app). ./reset-and-run.sh for a clean slate.
• Prerequisites: macOS 14.0+, Xcode, Apple Developer ID for code signing.
• Configuration: A bootstrap config is fetched from https://fazm.ai/api/bootstrap on first run, pointing to a hosted backend. Users can override this by creating a .env.app or .env.app.dev file in the repo root.
• Links: fazm.ai (website), GitHub asset links for demos (e.g., https://github.com/user-attachments/assets/42478070-9c32-48cc-b18a-61c646c6b6bd).

Highlighted Details

• Voice-controlled AI computer agent for macOS.
• Automates browser control, code writing, document handling, and Google Apps operations.
• Learns user workflows and offers "Smart Connections" across platforms.
• Supports "Visual Tasks" by understanding image context.
• Includes CRM management features for automated data entry.

Maintenance & Community

No specific details on contributors, sponsorships, or community channels (like Discord/Slack) are provided in the README.

Licensing & Compatibility

• License: MIT.
• Compatibility: The MIT license generally permits commercial use and linking with closed-source projects.

Limitations & Caveats

The project requires macOS 14.0+ and Xcode, limiting its use to newer Apple hardware. While described as "fully local," it fetches initial configuration from a hosted backend and requires runtime authentication for billable AI models, implying an internet connection is necessary for full functionality unless a custom backend is configured. The README does not detail the specific AI models used or their performance characteristics.