Your agent's tests are marking its own homework — read the article
Kun Chen banned Sonnet 5.5 from writing its own tests on DeepSWE and nothing got worse. Tests aren't dead, but a test written from the agent's own guess is no check at all.
Start here
JIM
DRURY.
Award-winning creative technologist, speaker, and digital strategist with over 15 years experience pushing the boundaries of design and technology.
I believe in building bold, breaking rules, and creating experiences that leave a lasting impact.
%3Afocal(657x692%3A658x693)&w=3840&q=75)
SpeakerCreatorStrategistTechnologist
Now
Head of Platform Innovation at Virgin Media O2 — shipping agentic engineering with humans in the loop.
Accolades
Storyblok
Storyblok Most Valued Professional 2026, Changemaker.
Storyblok
Storyblok Most Valued Professional 2025.
Storyblok
Innovation of the Year 2025 — Storyblok React Native Live.
Recent Writing
View all →
Devin dreams in markdown now, so somebody should review the dreams — read the article
Cognition put Devin's memory in a git repo of markdown and lets a nightly agent tidy it. Lovely primitive, but the writes are where memory goes wrong, so review them like code.
Uncle Bob stopped reading code. Fine, but who owns the instruments? — read the article
Uncle Bob doesn't read agent code anymore, he measures it. Fair enough, but the agents can edit the measurements, so decide who owns the instruments before you stop looking.