AI made implementation cheap. It made judgment scarce.
A working session for experienced engineers, run in person or online. We start with an ambiguous business problem and finish with production-quality software.
- Runs
- In person or online
- Shape
- Four working sessions
- Style
- Build alongside, not watch
- Bring
- A laptop and a real problem
What you leave with
Not a recording to work through later. A method you have already used once, on a problem that behaved like a real one.
Turn a vague request into a buildable plan
A repeatable way to move from “we need a dashboard” to a set of explicit product decisions, a domain model and a first vertical slice.
Ask the questions that change the product
How to find the small number of unanswered questions whose answers would materially alter what you build — and to stop asking the rest.
Give agents context they cannot misread
The artefacts that make coding agents reliable: decisions, invariants, conventions and a bounded slice. Delegate the change, never the ambiguity.
Tell a demo from a product
A production-readiness review you can run on any feature — authorization, idempotency, data, failure, deployment and observability.
How the day runs
One problem, carried the whole way through. Nobody watches me build.
- Session one
Don’t start with code
You get a real, deliberately under-specified business problem. We reverse the compression from feature request back to the behaviour someone actually wants to change.
- Session two
Scope and model
Cut to the smallest coherent workflow worth shipping. Name the concepts, the relationships and the invariants. Turn the workflow into state transitions.
- Session three
Build with agents
Architecture chosen in response to pressure rather than fashion. Then we implement vertical slices with coding agents, working from the decisions rather than around them.
- Session four
Make it production-grade
We review what we built the way an experienced engineer would: the invisible obligations that separate something that demos from something you can operate.
Cheap Code,
Scarce Judgment
The Engineer’s Guide to Building Products with AI.
- Seven chapters, 96 pages
- PDF and EPUB, both attached
- Free — it is the workshop’s thinking model
“Plausible software is not necessarily correct software.”
Where should I send it?
Tell me a little about where you are and I’ll send the book straight away.
Built for
- Senior and staff engineers who still build
- Tech leads and engineering managers close to delivery
- Teams shipping faster with AI but not more confidently
- Engineers who want to be useful earlier in the problem
Not built for
- Anyone looking for prompt tips or tool tricks
- Teams wanting a vendor-specific agent tutorial
- Engineers early enough that implementation is still the hard part
If your startup needs senior engineering judgment inside delivery, let's talk.
I work with founders and engineering leaders to improve technical foundations, reduce avoidable risk, and help teams ship critical product work with more confidence.