Skip to main content

Welcome to OpenHuman

OpenHuman is an open-source AI assistant designed to serve as your memory and actor across your tools. Built on Rust + Tauri and licensed under GNU GPL3, it bridges the gap between what AI models can do and what they actually know about you.

Every model in the world — over 200 of them — has the same fundamental limitation: they are stateless. You input a prompt, get a response, and the context is gone. Even those with "memory" only store a few bullet points. A few bullet points are a sticky note, not intelligence.

OpenHuman addresses this with a deliberately different stack:

  • Local-first Memory Tree. Every connected source. Gmail, Slack, GitHub, Notion, your own notes — processed through deterministic pipelines: normalized Markdown, ≤3k-token chunks, scoring, collapsing into source/topic/day summary trees. Stored in SQLite on your machine. Not a vector soup black box.
  • Obsidian-style wiki. Existing as .md files in a vault you can open, browse, edit, and manually link in Obsidian. Inspired by Karpathy's obsidian-wiki workflow. You can't trust memory you can't read.
  • 118+ third-party integrations. One-click OAuth for Gmail, GitHub, Slack, Notion, Stripe, Calendar, Drive, Linear, Jira, and more — no manual API key connections, no navigating plugin marketplaces.
  • Auto-fetch. Every twenty minutes, OpenHuman pulls new data from each active connection and collapses it into your memory tree without you asking, so the agent already has tomorrow's context in the morning.
  • Agents built for big data. Smart token compression (TokenJuice) compresses verbose tool outputs before they enter the LLM context, so scanning six months of email costs only single-digit dollars. Automatic model routing sends each task to the right model — hint:reasoning to frontier models, hint:fast to cheap models, vision to vision — all under one subscription. Optional local AI via Ollama or LM Studio keeps supported workloads on-device.
  • Battery included. Full agent toolbelt with default connections: web search, web scraper, full code toolkit (filesystem, git, lint, test, grep), browser and computer control, scheduling, memory tools, agent coordination for spawning sub-agents, and native voice — STT input, TTS output, mascot lip-sync, and a live Google Meet agent that joins meetings, transcribes to the memory tree, and can call back. No "install a plugin to read a file" hassle.
  • Simple, UI-first. A clean desktop experience and short onboarding path get you from installation to a working agent in a few clicks — no configuration-first setup, no terminal. Agents have faces: a talking desktop mascot that reacts to its surroundings, joins your Google Meets as a real participant, remembers you across weeks, and keeps thinking in the background when you stop typing.

Together, these transform OpenHuman into something fundamentally different from a chatbot. It's an AI agent that consumes vast amounts of personal data at low cost, maintains a persistent and evolving understanding of your world, and takes proactive action on your behalf.

Note: OpenHuman is not AGI. But it is a meaningful architectural step toward better memory, better orchestration, and better tools.

Core Features​

FeatureDescription
Memory TreeLocal SQLite knowledge base, automatic ingestion from all connected services
Obsidian SyncMemory as Markdown files, editable directly in Obsidian
118+ IntegrationsOne-click OAuth for Gmail, Slack, Notion, GitHub, and more
Auto-fetchAutomatic sync of new data every 20 minutes
Token CompressionSmart compression of tool outputs, cost control
Model RoutingAutomatic selection of the best model for each task
Local AIOptional Ollama/LM Studio, privacy-first
Native VoiceSTT + TTS, Google Meet meeting assistant

Quick Start​

  1. Install OpenHuman
  2. Complete app onboarding
  3. Connect your first service
  4. Make your first request

Join the Community​