Focus

Agents

Multi-step systems where the hard part is handoffs, state, and failing closed — not a clever prompt.

Agents fail at orchestration, not reasoning. A chain of 90% steps becomes a 59% system unless every boundary is boring and checkable.

This page collects agent work: schema gates between tools, speculative multi-agent setups, graph-augmented retrieval, and compliance copilots that explain their rules. More builds will land here as they ship.

On this shelf

← Back to the letter