Test & Learn

Turn AI experiments into useful decisions.

Follow the build, see what the checks reveal, and take away a practical next step. Test & Learn helps product leaders and teams decide what to try, what to improve, and where more evidence is needed.

Your Team Bought AI Coding Tools. It Still Needs an Operating System.

The hard part is no longer getting access to an AI coding assistant. It is designing the decisions, context, checks, and ownership that turn fast output into dependable work.

Read the field note

Builds, protocols, and operating choices.

New entries are published only when there is something inspectable to show or a useful question to sharpen.

01
Built

The agentic coding operating system

A practical model for moving from individual tool access to shared, reviewable team practice.

02
Exploring

A2A beyond the protocol diagram

What agent-to-agent communication changes—and does not change—for product and enterprise teams.

03
Built

Trusted agents need evidence

A lightweight evidence ladder for separating experiments, artifacts, tests, deployments, and outcomes.