OpenHuman is a personal AI system written in Rust that functions as a local-first alternative to cloud-based AI assistants. The project describes itself as a brain that builds persistent memory of your life, an orchestrator of agent fleets and workflows, and a deep researcher. The repository has achieved significant traction since launch, becoming the number one trending repository on GitHub for nine consecutive days within its first week.
The system operates around three core pillars. The brain component includes a Memory Tree that stores compressed Markdown data in SQLite on the user's machine, mirrored as an Obsidian vault for direct editing and review. This approach avoids the black-box nature of vector databases. The brain integrates with over 100 OAuth services, 5,000 MCP servers, and 90,000 skills, enabling one-click connections to Gmail, Notion, GitHub, Slack, and other tools. An auto-fetch feature updates the memory every 20 minutes. The system includes a subconscious background loop that continuously diffs the user's world, advances goals, and generates morning briefings. TokenJuice compresses tool output before sending it to language models, reducing token usage by up to 80 percent.
The orchestrator component manages agent fleets through workflows that users can review and save on a canvas interface. It uses open-source tinyflows and tinyagents libraries to run durable, trigger-driven, approval-gated automation. The system features a split-brain architecture with a fast reflex agent handling incoming traffic while a deep reasoning core delegates to worker fleets. An agent economy layer provides Signal-encrypted agent-to-agent orchestration through tiny.place with USDC bounties and trading capabilities.
The deep researcher and doer capabilities include SuperContext, which scans memory and files before processing user messages. The system provides web search, scraping, coding tools, a native browser, and in-process Whisper voice processing. Model routing selects appropriate language models per task across a single subscription, with optional local AI support. Meeting agents can join Google Meet, Zoom, Teams, and Webex calls with voice and video, auto-joining from calendars and generating summaries with action items. Image and video generation through Seedream and Veo integrate directly into the workspace.
The repository maintains active development with 171 open issues as of the latest tracking period, showing growth of 2 issues from the previous check. GitGenius tracking reveals median issue and pull request response latency of 0.0 hours with a mean of 2.6 hours across 1,541 items, indicating rapid community engagement. The most active issue labels are bug (205), feature (144), and agent (116). Primary contributors include senamakel with 2,308 tracked events, graycyrus with 979 events, and oxoxDev with 582 events. The project overlaps contributors with github/gh-aw, solo-io/gloo, and longhorn/longhorn repositories.
OpenHuman emphasizes privacy and security through on-device encrypted data, approval gates, OS-keyring secrets, and optional sandboxing. A Privacy Mode switch prevents any inference from leaving the user's machine, enforced at the Rust core level. The system supports 17 messaging channels including Telegram, Discord, Slack, WhatsApp, Signal, iMessage, and native email with IMAP IDLE and SMTP. The interface prioritizes simplicity with no configuration files required, featuring a mascot character that speaks, reacts, and remembers users. Customization options include five theme families and a full visual theme editor with JSON export capabilities.
