Model taste
Most AI disappointment starts with weak models, weak context, or weak expectations.
Jacob / AI field notes
AI is moving faster than corporate comfort.
Sharp notes, practical arguments, and working models for teams that need to turn AI from nervous slideware into shipped leverage.
5 essays
4 adoption levers
1 bias: ship
Signal
Not hype. Not fear. Useful pressure.
I am a Chapter Lead Software Engineering at the 5th largest firm in New Zealand, with almost a decade as a software engineer and real experience in end-to-end delivery of production-grade systems. I care about the practical edge of AI: where it helps teams think faster, build cleaner systems, and become more ambitious without losing discipline.
This site is intentionally opinionated. AI is not just another tool rollout. It changes taste, leverage, feedback loops, and the shape of software work itself.
Posts
Five arguments and one AI lab.
Corporate AI rollout simulator
Tune token usage, security stance, context quality, and workflow integration to see how readiness, cost, risk, and adoption move.
AI is soooo quick: how to get on the train
A starter guide for early-career people, plus an interactive quiz that turns your current confidence into a learning path.
The AI Divide: why people love it, hate it, and miss it
AI is often judged by breaches, sameness, poor models, and first-day frustration. The real question is whether people learn the instrument.
How to get corporates ready for AI
Security culture, frontrunners, paved roads, and the boring operating model that makes AI adoption real.
The future of software engineering
Software work splits into builders and critical application maintainers. The maintenance-only lane is the one under pressure.
An homage to Karpathy
Recognition for the person who gave many of us language for this shift, plus two AI supporters: AutoResearch and Second Brain.