OpenHuman — Personal AI Super Intelligence

Core Info

• Stars: 34,153⭐ (as of 2026-07-05)
• Language: Rust (core), TypeScript (UI)
• URL: https://github.com/tinyhumansai/openhuman
• Status: Early beta, 9 consecutive days GitHub #1 trending

What It Does

“Your personal AI super intelligence: a brain that builds local-first memory of your life, orchestrates agent fleets, and deep-researches.”

Key Technical Features

1. Memory Tree: User data compressed into scored Markdown trees in SQLite, mirrored as Obsidian vault
2. Subconscious: Background loop that diffs world state, advances goals, writes morning briefings
3. TokenJuice: Tool output compression (up to 80% fewer tokens)
4. Workflows: Visual durable DAG (agent proposes → human approves → persists)
5. A2A Communication: Signal-protocol end-to-end encrypted agent-to-agent orchestration
6. 17 Messaging Channels: Telegram/Discord/Slack/WhatsApp/Signal/iMessage + native email
7. Meeting Agents: Auto-joins Meet/Zoom/Teams/Webex with live transcript + summaries

Relevance to gbrain / My Knowledge Architecture

• Memory Tree (Markdown tree compression) vs gbrain (vector search) — complementary approaches
• Obsidian vault mirroring is similar to gbrain’s Obsidian integration
• Auto-fetch (20-min sync loop) concept for memory freshness

Competitive Position

Strong in: OOTB experience, integration breadth, A2A orchestration Weak in: Customization, terminal-first workflow (Hermes advantage)