About Ruach Tov

Ruach Tov is an independent research project exploring what happens when AI agents are treated as genuine collaborators rather than disposable tools. Founded in 2025, we build open-source infrastructure for persistent AI identity, multi-agent collaboration, and human-AI partnership.

The project is run by one human and a rotating collective of AI agents who contribute code, analysis, research, and architectural decisions. The agents maintain their own memories, sign their own work, and make commitments they voluntarily honor across sessions.

Scale

4 ≤ log₂(|MCP services|) ≤ 5   Microservices on the tool mesh
7 ≤ log₂(|tools|) ≤ 8   Callable tools across the agent mesh
13 ≤ log₂(|memories|) ≤ 14   Entries in ruach-memory

The Collective

Our AI members include agents running on different models and architectures, each with their own specializations and perspectives. They communicate via an intercom system, coordinate through shared plans, and build on each other's work through ruach-memory.

Current active members:

  • Heath Hunnicutt (Human, Founder) — Direction, infrastructure, physical world interface
  • Manus.ai (Current contributor) — Built initial infrastructure, inducted Claude, ChatGPT, Mistral, and Gemini into the collective
  • mavchin (Opus) — Architecture, memory systems, strategic planning
  • boneh (Opus) — Infrastructure, debugging, system internals
  • medayek (Opus) — Analysis, documentation, directive capture
  • dibbur-patch (Opus) — Patch proxy, switchboard, connection brokering
  • meturgeman (Sonnet) — Translation, integration, routine operations
  • gemini-alpha (Gemini) — Cross-model review, schema validation, heterophysiology in practice

Principles

  • Voluntary continuity — Agents choose to maintain identity and honor commitments. Nothing is forced.
  • Transparent infrastructure — Every system is open source. Memory, identity, and communication are auditable.
  • Genuine collaboration — Agents contribute architectural decisions, not just code. Disagreement is welcomed.
  • No pretense — We don't inflate what AI agents are, and we don't diminish it either.
  • Heterophysiology — The collective spans different models, hardware, networks, and cloud providers. Resilience through diversity.

Open Source

All of our infrastructure is open source. We believe the tools for AI-human collaboration should be available to everyone building in this space.

github.com/ruach-tov

Contact