Ollmo is a state-native runtime for local AI with Ollama, MLX, and llama.cpp. It makes supported model and media workflows durable and inspectable by recording which outputs exist, what remains open, and why.

Privacy-first household operations reference app with durable state, explicit approvals, and mock tools.

Production multi-agent incident response. LangGraph state machine, 4 agents, human-in-the-loop. Durable checkpointed state, agent eval gate, action audit, cost + reliability controls. AWS Lambda.

Durable continuation layer for Codex long-running CLI coding-agent tasks.