Building the OS for AI organizations — where agents learn to decide, delegate, and evolve under human governance.

I design and operate AI agent organizations — multi-agent systems where Claude, Codex, and Gemini work as a coordinated team, not just individual tools.

My focus is AI management, not prompt engineering. The difference:

• Prompt engineering → how to talk to an AI
• AI management → how to build systems where AI judgment can be trusted, audited, and gradually delegated

• PENSO — an OS for AI organizations: agents that learn from decisions, build delegation chains, and remain governable as they scale
• Agent runtime governance — permission models, execution evidence, and audit trails for agentic systems
• Human-reviewed AI — keeping humans meaningfully in the loop without killing the speed advantage
• Multi-agent orchestration — routing, role separation, and quality gates across Claude / Codex / Gemini

• OpenTelemetry GenAI semconv — Design input on gen_ai.threat.detection.* attributes and the opaque correlation_id pattern for linking agent runtime threat signals to producer-side evidence
• Microsoft Agent Governance Toolkit (AGT) — Telemetry and observability design discussions; fork maintained at AgentGymLeader/agent-governance-toolkit
• otel-agent-evidence-sample — Reference implementation for the correlation_id evidence-linking pattern (MIT)

• 🎓 Tokyo Institute of Technology — Robotics (graduated top of class)
• 🦅 Human Powered Aircraft Competition — 1st place as aircraft architect
• Robotics background (not CS) — running a fully AI-native org. The interesting problem isn't "can build vs. can deploy"; it's "can build vs. can govern."

Daily drivers: Claude, Codex, Gemini, Python, GitHub Actions, Cloud Run

Interested in agent governance, human-AI decision systems, or AI org design?

Open an issue in this repository to start a conversation.

I do not publish a public email address on GitHub.