-
Save the Boring Time
The biggest AI time-save is not writing features, it is the ops work that surrounds them. Three flows, two automation surfaces, and the one rule that keeps it safe.
-
Pick a Framework, Any Framework
AI-DLC is one option, not the option. Four methodologies in 2026, the foundation they all share, and the worktrees-and-orchestration upgrade that scales any of them.
-
Context Is Not a Dev Job
Half the context an AI needs lives in the heads of the people outside engineering who own intent. AI-DLC only works when that half gets written down, in the same vault and the same spec template the engineers use.
-
Trust the Pipeline
Every 'we need to ship faster' conversation is actually a 'we do not trust the pipeline' conversation. Five properties of a trusted CI, plus the AI-written Playwright and Testkube setup that gets you there.
-
Your AI Context Belongs to the Team
Every session with an AI is a teaching session. Most teams throw the teaching away. Here is how to keep it, share it, and let it compound.
-
You Cannot Push a Developer
Adopting AWS AI-DLC across an engineering org taught me the verb was wrong. You do not push people toward a methodology. You remove the reasons to say no.
-
Frictionless: The Book Behind the Loop
Nicole Forsgren and Abi Noda wrote the book on removing developer friction. The spec-driven loop from last post is one small, stubborn instance of it.
