OpenHuman

OpenHuman is an open-source agentic assistant designed to integrate with you in your daily life. It combines local memory storage, managed services where needed, with a simple yet extremely powerful interface. Think of it as your personal AI superintelligence that learns from your work, remembers your preferences, and helps you accomplish tasks more efficiently.

Unlike other AI assistants, OpenHuman offers: • Local-first memory with Obsidian-style vaults • 118+ third-party integrations with auto-fetch • Smart token compression (TokenJuice) • Persistent memory that remembers you across sessions • Simple UI-first experience with no terminal setup required • Open-source GNU license with full transparency

OpenHuman currently supports: • macOS (via Homebrew) • Linux (Debian/Ubuntu via apt, Arch via AUR) • Windows (via MSI installer) We're working on expanding to more platforms based on community demand.

Installation varies by platform:

macOS: brew tap tinyhumansai/core && brew install openhuman
Linux (Debian/Ubuntu): Follow our apt repository setup
Windows: Download the signed .msi installer
Manual: Download platform-specific installers from our releases page

Yes! OpenHuman is open-source and free to use. The core application is available under the GNU General Public License v3.0. We offer optional premium features and model subscriptions for enhanced capabilities, but the base functionality is completely free.

Minimum requirements: • 8GB RAM (16GB recommended) • 2GB free disk space • Modern CPU with SSE4.2 support • Stable internet connection for initial setup • macOS 10.14+, Ubuntu 18.04+, Windows 10+

The Memory Tree creates a local-first knowledge base from your activities. It automatically: • Canonicalizes data into ≤3k-token Markdown chunks • Scores and structures information hierarchically • Stores everything locally in SQLite • Creates an Obsidian-compatible vault for browsing/editing • Compresses context using TokenJuice technology

OpenHuman supports 118+ integrations including: • Communication: Slack, Discord, Telegram • Productivity: Notion, Trello, Asana • Development: GitHub, GitLab, Jira • Cloud: Google Drive, Dropbox, OneDrive • Services: Gmail, Calendar, Stripe, Linear All integrations use one-click OAuth and auto-fetch data every 20 minutes.

Yes! OpenHuman prioritizes local-first operation: • Memory Tree and workspace config are always local • Basic functionality works offline • Some features require internet (model routing, web search) • Optional local AI via Ollama for supported workloads

Your data is stored locally by default: • Memory Tree and Obsidian vault on your machine • Runtime state encrypted locally • Managed backend only handles account auth and model routing • You can bring your own credentials for full privacy

OpenHuman uses multiple security layers: • Local encryption of sensitive data • Zero-knowledge architecture where possible • Secure OAuth flows for integrations • Regular security audits • Transparent open-source codebase

Absolutely not. OpenHuman is committed to privacy: • Your data belongs to you • No data mining or profiling • No third-party data sharing • All processing happens locally when possible • Clear opt-in for all cloud features

OpenHuman is built with: • Rust for core performance and safety • JavaScript/TypeScript for UI (React-based) • Python for integration connectors • C++ for native modules The architecture emphasizes performance, security, and cross-platform compatibility.

Yes! OpenHuman supports self-hosting: • Local memory backend (SQLite) • Self-hosted models via Ollama • Custom Composio credentials • Optional backend components Full documentation available in our GitBook.

TokenJuice is our smart compression system that: • Converts HTML to Markdown • Shortens long URLs • Deduplicates and summarizes verbose output • Preserves CJK, emoji, and multi-byte text • Reduces costs/latency by up to 80%

Try these solutions: • Use the recommended native package installers • Ensure your system meets requirements • Check for conflicting software • Try manual installation from releases • Visit our Discord for community support

Common fixes: • Re-authenticate the integration • Check API permissions • Verify service status • Clear cache and restart • Check our integration documentation

We welcome community contributions: • Report bugs on GitHub Issues • Use the template provided • Include system information • Request features via GitHub Discussions • Join our Discord for direct feedback