Building governed agent teams
How TeamOS separates specialist authority, preserves evidence, and returns failed work to the right stage.
BLZ Research / Field notes
BLZ Research is the learning arm of the Foundry. It captures the decisions, failures, controls, and reusable patterns behind active work—so every build begins with better questions and leaves behind useful evidence.
Published when the underlying work is clear enough to be useful—not merely because content is due.
How TeamOS separates specialist authority, preserves evidence, and returns failed work to the right stage.
Lessons from AI Build Crew on workload modeling, visible economics, evaluation, and human approval.
Why a convincing prototype can still be the wrong product when positioning, audience, or identity is unresolved.
Research rule