Start here
The board came back. The highlights lied. (opens in new tab)Codenames AI →Articles
TEAM Portability
Field reports from rebuilding how I work with AI. Taste, Evals & Verification, Authority, and Measurement became four recurring disciplines. Portability is what makes those practices transferable across projects, tools, and agents.
Start here · Six-month retrospective
TEAM Portability: Six Months of Rebuilding How I Work with AI
How 21 field reports became a framework for engineering with AI, and what six months of experimentation changed about the way I work.
Taste
What is worth building, and what counts as a good outcome?
5 reports
Coming back should restore what the product treats as current truth, not whatever was on screen.
Evals & Verification
Does the implementation satisfy its expected contract?
4 reports
Passing the check is not the same as getting the draft right.
Start here
I fixed my AI reviewer. Then I kept solving the wrong problem (opens in new tab)Editorial workflow →Authority
What are agents permitted to change?
5 reports
You still have to define where agent work stops.
Start here
The agent host didn't have the lifecycle boundaries I assumed (opens in new tab)Measurement
What does observed system behaviour reveal about quality and outcomes?
3 reports
A metric is only useful when you know what it counts.
Start here
Active players looked real until we asked which sessions counted (opens in new tab)Codenames AI →Portability
How do we apply these practices consistently across projects, tools, and agents?
4 reports
Copying agent setup is easy; keeping it aligned across projects is not.
Start here
I was solving agent portability at the wrong boundary (opens in new tab)Every report is published on DEV. Full archive ↗