Concept

Provenance Engineering

Recording who decided what, when and on what evidence — an unbroken chain of custody for decisions in human-plus-agent systems.

In a co-produced system, "who decided this?" stops being easy: a single decision may pass through a human prompt, retrieved documents, model inference, agent tool calls, model review and human approval. Provenance engineering treats provenance not as a log written after the incident but as a typed, signed predicate emitted when the decision is made — an unbroken chain of custody.

It is where accountability, debugging and compliance converge: debugging a co-produced system is, at bottom, a provenance query. Depth: Provenance Engineering: Reconstructing Who Decided What When Humans, Agents and Models All Contributed.