I look after engineering by day, but the part of the work that's grabbed me lately happens in the margins: using AI, almost every day, to build and test new product ideas.
It has changed the shape of how I prototype. The distance between "what if" and "something I can actually click on" used to be measured in weeks. Now it's an afternoon — sometimes an hour.
What it's good at
- Killing ideas faster. The point of a quick build isn't to ship it; it's to find out the idea was wrong before you spend a month on it. AI makes that loop cheap.
- A patient pair. It doesn't get tired of the boring scaffolding — the test data, the tenth variation of a prompt or a schema.
- Range over depth. It'll get you 80% of the way into an unfamiliar domain astonishingly fast. The last 20% is exactly where your own judgement has to take over.
What still humbles me
It is confidently wrong often enough that you can't outsource the thinking. The skill that matters more than ever is the old engineering instinct: assume nothing, verify everything. AI makes building faster; it makes verifying more important, not less.
Tinkering, and learning it properly
The daily play is one half; the other is going and learning the foundations properly so the instinct keeps up with the tooling. I did the Post Graduate Program in AI & Machine Learning at UT Austin, then the AI Engineering Accelerator at Outskill, and I'm currently in their AI Engineering Fellowship — agents, frameworks, multi-agent systems, MCP, the works. The two halves feed each other: the fellowship sharpens what the tinkering stumbles into.
More of these as I go — less hype, more notes from actually using the thing.