Latest
Posts
Writing With AI Without Losing The Plot
How useful AI writing starts with context, taste, and a clear reason to exist.
Why Agentic Delivery Needs Receipts
Logs, screenshots, diffs, and checks make AI-assisted work inspectable instead of mysterious.
From Debugging Incidents To Operating Memory
Repeated production lessons become reusable operating patterns when agents preserve the right evidence.
Shipping With AI Without Losing Human Taste
The review loops that keep speed from turning into generic output.
What We Learned From Ten Verification Loops
A practical look at build checks, browser proof, and the point where a bug becomes an architecture problem.
Turning Client Context Into Agent Instructions
How product notes become constraints, workflows, and reusable delivery memory without leaking private details.
The Quiet Work Of Not Shipping Secrets
A field note on credential boundaries, safe previews, and why secret handling should be boring.
How Agents And Humans Divide The Work
Agents do the work, humans make the decisions. This is what that looks like inside a real delivery team.