OK, so after chating with @_lopopolo @GeoffreyHuntley @sschillace and @rwitoff what I really want to see is examples of the deterministic quality gates that are rigorous enough to avoid sprawl when you go full agent mode on your prod code base. Esp for architecture and
Helping CTOs scale AI adoption. Hands-on agentic engineer. Author @OReillyMedia. CTO communities @GatherDev. Exec summits @Kubecon_ Taught DS @Columbia_Biz
- This works, but you HAVE to have your deterministic quality gates nailed. I didn't and ended up with a sprawling set of "fixes" I had to send fable to rip out and simplify. Context and adversarial reviews are just suggestions :(A weird experiment I've been trying the last few weeks is having Claude take over day-to-day maintenance of our apps. Seeing early signs of life that this might be possible. The setup is straightforward: we have a Slack channel called proj-claude-maintains-apps. In it, Claude
- Anyone else notice that when they run agents interactively at a higher level of intelligence, sometimes they will take responsibility for "human" In the loop steps? Can't decide if they are stepping into their power or I simply prompted poorly :)
- This feels more and more like managing a team. I feel like my agents are finally through the forming and storming. Now landing on the context required for norming and finally really seeing some performing :)
- Anyone else have occasional days with their agents where they think we should just be chiseling hieroglyphs in stone tablets because it might be slow but at least it would be unambiguous? #agentlife


