The Haunt.
Short reads on software in the age of AI. 3 minutes or less. Come back often.
- Building with AI
A code review prompt that starts from the request.
Give the reviewer the request, not just the diff, and treat what it says as leads to check.
3 min read - Building with AI
Seven AI coding guidelines a team will actually follow.
Write rules for the people first, then tell the agent what the people decided.
3 min read - Building with AI
You don't need a new tool to write test cases with AI.
A requirement and a chat assistant are enough, and the skill is choosing which cases matter.
3 min read - The craft
Stop resubmitting. Debug the failing case.
Each resubmit without a reason is a guess, and the failing case is telling you exactly where to look.
3 min read - Building with AI
Stop re-prompting. Debug it.
Asking the assistant to try again is a guess, and debugging is how you stop guessing.
3 min read - The craft
Flaky tests need a method, not a retry button.
A flaky test is a bug that only shows up sometimes, so debug it like one.
3 min read - Building with AI
Introduce AI to your team with an hour, not a slide deck.
People trust a tool they have caught making a mistake, not one they watched in a demo.
3 min read - Building with AI
Your AI tests pass. Do they check anything?
Mutation testing is "make every test fail once", done by a machine.
3 min read - Staying relevant
Will AI replace QA? It is making more to check.
When code is cheap to produce, the people who can prove it works become the bottleneck.
3 min read - Building with AI
Can AI write your unit tests? Only if they can fail.
A test you have never seen fail is a test you are trusting, not checking.
3 min read - The industry
Measure what AI changes, not how much it writes.
Count what reaches users and stays there, not lines generated.
3 min read - Building with AI
How to review AI-generated code.
Start from what was asked, not from the tidy code you were handed.
3 min read - Staying relevant
Use AI as a tutor, not a ghostwriter.
The same tool can teach you or think for you, and the difference is how you ask.
3 min read - The craft
Read more code than you write.
When code is cheap to produce, the scarce skill is understanding it.
2 min read - The industry
Writing code was never the whole job.
As implementation gets faster, the questions around it become harder to ignore.
2 min read - Building with AI
The hard part starts after the demo.
A convincing prototype and a dependable product ask different things of you.
2 min read - Staying relevant
You don’t need to learn everything.
Build a learning practice that survives the next framework, model, and headline.
2 min read