Field notes on AI in production.
What we've learned shipping AI systems that have to work when it counts. No hype, no theory — the engineering reality.
Why most Claude prototypes never reach production
The demonstration succeeds, yet the system fails under real use. An examination of the gap most teams never scope for.
Read article →Why most Claude prototypes never reach production
The demonstration succeeds, yet the system fails under real use. An examination of the gap most teams never scope for.
May 2026Practice · 5 minEvaluations: the difference between a demonstration and a product
Without a way to measure whether your AI is improving or regressing, every release is a guess. How we establish evaluation suites.
Apr 2026Economics · 4 minFractional versus full-time: the real cost of an AI hire
A $200k salary, months of recruiting, and ramp time, weighed against senior engineering delivered in weeks.
Apr 2026Engineering · 7 minObservability for LLM systems: seeing what your model actually does
You cannot operate what you cannot see. The traces, logs, and dashboards that make an AI system accountable.
Mar 2026Security · 5 minGuardrails that hold: handling the inputs you didn't plan for
Real users do unexpected things. Designing for the long tail of inputs is most of the work of going to production.
Mar 2026Practice · 6 minKeeping AI costs predictable as usage grows
Token budgets, caching, and model routing. The unglamorous engineering that keeps unit economics from drifting.
Feb 2026