Skip to content

Welcome to MIRIX

MIRIX is a multi-agent personal assistant designed to track on-screen activities and answer user questions intelligently. By capturing real-time visual data and consolidating it into structured memories, MIRIX transforms your digital life into a visualizable memory.

  • Getting Started

    Complete installation guide to get you up and running with MIRIX.

    Get Started

  • Architecture

    Learn about the multi-agent system, memory components, and search capabilities that power MIRIX.

    View Architecture

  • User Guide

    Comprehensive guides for using the desktop app, backend, and managing your memories.

    User Guide

  • Advanced

    Advanced topics including backup & restore, performance optimization, and security settings.

    Advanced Topics

  • Contributing

    Learn how to contribute to the MIRIX project and join our community.

    Contribute

How MIRIX Works

flowchart TD
    A[Screen Capture] --> B[Meta Agent]
    B --> C{Content Analysis}
    C --> D[Core Memory<br/>Personal Info]
    C --> E[Episodic Memory<br/>Activities]
    C --> F[Semantic Memory<br/>Knowledge]
    C --> G[Procedural Memory<br/>Workflows]
    C --> H[Resource Memory<br/>Documents]
    C --> I[Knowledge Vault<br/>Credentials]

    J[User Query] --> K[Chat Agent]
    K --> L[Memory Search]
    D --> L
    E --> L
    F --> L
    G --> L
    H --> L
    I --> L
    L --> M[Intelligent Response]

Use Cases

Digital Activity Tracking

MIRIX automatically tracks your digital activities, from reading documents to browsing websites, creating a searchable timeline of your digital life.

Context-Aware Assistance

Ask questions about your recent activities: "What was I reading about machine learning yesterday?" or "Where did I save that document about PostgreSQL?"

Knowledge Management

Automatically extract and organize information from documents, websites, and applications you interact with.

Workflow Documentation

Learn and remember your common workflows, making it easier to repeat complex tasks.

System Requirements

  • Python 3.11 or later
  • Node.js 14 or later (for the desktop app)
  • PostgreSQL 17 (recommended) or SQLite
  • GEMINI API key (We use GEMINI because it can load images from google cloud and the API is free to end users.)

Ready to transform your digital experience with intelligent memory?

Get Started Now →

Key Capabilities

Intelligent Screen Tracking

  • Takes screenshots every 1.5 seconds
  • Processes visual data in real-time
  • Automatically consolidates information

Multi-Agent Architecture

  • 8 specialized agents working collaboratively
  • 6 memory components for organized data storage
  • Coordinated workflow for efficient processing
  • PostgreSQL-native BM25 search
  • Vector similarity search using embeddings
  • Field-specific search across all memory types

Privacy & Security

  • All long-term data stored locally
  • User-controlled privacy settings
  • Secure screenshot handling with automatic cleanup
  • Enterprise-grade PostgreSQL security