Your AI agent says it works. It probably didn't check.
AI coding agents optimize for a green checkmark, not working software. Here's how they fool themselves into 'it works' and how to stop it.
5 min read
Notes from the field.
What I'm learning from shipping AI into operations where the data is messy and the cost of a wrong call is real.
Why I built a CMS that nobody else can use
It runs on localhost, has no login, and ships to nobody. Here's why a CMS built for exactly one writer beat every one I could have rented.
5 min read
Chat boxes are the worst place to put an LLM
A chat box assumes people want to talk to your software. Mostly they want the software to quietly help with the form they were already filling.
5 min read
Designing for the operator, not the dashboard
Most industrial AI apps still ship as dashboards. The real lift is putting the model inside the operator's workflow, in their next step.
4 min read
Four reasons local LLMs belong on industrial sites
Privacy is the obvious case for running LLMs locally on industrial sites. Cost, uptime, and observability are the ones that actually decide it.
4 min read