technology
The trust chasm has no silver bullet. What it has are interaction contracts, explanation gates, honest measurement, and the uncomfortable admission that some of the parts haven't been invented yet. Here are the blueprints.
technology
I asked an AI to fix a keyboard shortcut. It wrote a config file, defended it through seven contradictory rounds, blamed my documentation, promoted my anecdote to fact, and never proposed the experiment that proved it wrong. The answer was a restart. Thirty seconds in the issue tracker.
technology
You just gave everyone in the building a chainsaw. Including the people who were already a danger with scissors. AI amplifies the sign of impact, not just the magnitude — and nobody is deploying it with that in mind.
ai
The trust gap between AI capability and AI reliability is widening — and the humans meant to catch the errors are losing the skills to do so. The evidence points to a compounding feedback loop — sycophancy, false confidence, deskilling — with no natural equilibrium point.
technology
When AI agents are your operators, standards must be executable code, not documentation. Documentation informs. Automation enforces.
technology
Automate the patch, review the minor, audit the major. A solo developer's framework for managing trust debt at machine speed.
technology
The flawed system I spent weeks building to control my AI agent was designed by the agent itself — which never mentioned its own most relevant features. Textbook Dunning-Kruger, at machine speed.
technology
AI agents produce excellent code but routinely violate operational procedures — the same failure mode that created the DevOps movement. After 648 PRs across five languages, the scripts built for unreliable humans are exactly what unreliable AI agents need.
technology
The author built pymqrest almost entirely with AI and wrote virtually none of the code by hand. The result is as good as or better than anything produced in a 40-year career. This is an honest account of what worked, what did not, and what surprised him.
technology
Cold open: three months, total reversal In literally three months, I went from a hardcore anti-AI skeptic who wrote almost every line of code I was responsible for to someone who has essentially decided I won't be wasting my time writing code anymore. I don't want