More Help
Setup details: install guide. Official facts: GitHub README and docs. Bugs: official issue tracker.
Still Have Questions?
If your question is not answered above, check the detailed guides or reach out. The install guide covers platform-specific steps. The privacy guide explains data flows before you connect accounts. The setup guide walks through first-launch configuration.
What is OpenHuman?
An open-source, desktop-first personal AI agent built by TinyHumans. It syncs your email, calendar, repos, notes, and chats roughly every 20 minutes, builds a local memory tree, and exposes that memory to an assistant that acts through 118+ OAuth integrations.
Is OpenHuman open source?
Yes. The source is on GitHub under GPL-3.0. Check the repository license file and current state before relying on this for compliance decisions.
Is OpenHuman safe to install?
Safety depends on install source, OAuth scopes, and integrations. Use the signed installer, read scripts before running them, start with a low-risk test account, and wait for an independent security audit before connecting highly sensitive accounts.
Does OpenHuman run locally?
Local-first, not local-only. Memory, OAuth tokens after the initial exchange, embeddings, and summary-tree building stay local. Chat, reasoning, vision, ElevenLabs TTS, web search, and Google Meet route to cloud providers by default.
Is it free?
GPL-3.0 and free to install. A single subscription unlocks 30+ AI providers; verify current pricing on tinyhumans.ai/openhuman. Route through Ollama or LM Studio to skip the subscription for supported workloads.
What operating systems are supported?
macOS, Windows, and Linux. macOS builds are available as DMG for Apple Silicon and x64. Windows builds are available as EXE. Linux builds support x86_64 and aarch64, with AppImage as an alternative.
How many integrations does it have?
118+ integrations via one-click OAuth: Gmail, Calendar, Drive, Notion, GitHub, Slack, Stripe, Linear, Jira, Telegram, Discord, Zoom, Outlook, Dropbox, and more.
What is the Memory Tree?
A hierarchical, summarized memory layer. Data converts to Markdown, gets chunked, scored, and folded into per-source, per-topic, per-day summaries in local SQLite and an Obsidian-compatible vault you can open and edit.
Can I share memory with Claude Code, Cursor, Codex, or OpenCode?
Yes. Set memory.backend to agentmemory in config.toml. The same durable store backs multiple coding agents simultaneously. See the local-ai page for setup notes.
Does it work offline?
Partially. Memory embeddings, summary-tree building, and reflective loops run locally with Ollama or LM Studio. Heavy reasoning, vision, voice TTS, and web search still route to cloud providers unless you configure around them.
Is there a security audit?
Not yet. The team has committed to an independent audit, but nothing is published. Empty CVE history reflects the project's youth, not proven security.
Where do I report bugs?
File at github.com/tinyhumansai/openhuman/issues with your OS, architecture, app version, failing connector, log lines, and whether local AI is enabled.
How does the Memory Tree work in detail?
The Memory Tree has three layers: themes (work, family, finance), entities (people, companies, repos), and documents (emails, notes, transactions). Data flows through fetch, canonicalize, chunk, summarize, and store steps. Everything lives in local SQLite and an Obsidian-compatible Markdown vault. See the openhuman-memory-tree page for a full deep dive.
Which integrations are available?
OpenHuman supports 118+ integrations across communication (Gmail, Slack, Discord, Telegram), productivity (Drive, Notion, Calendar), development (GitHub, GitLab), and business (Stripe, HubSpot) categories. See the openhuman-integrations page for the full list and security best practices.
What should I do after installing OpenHuman?
After installation, verify the app launches correctly, set your Memory Tree vault path, connect one low-risk integration, configure model routing, and optionally set up local AI with Ollama or LM Studio. See the openhuman-setup-guide page for a complete post-install checklist.