Writing
What we're learning
Short essays on what holds up when AI does real work, and shorter notes on what is being read and tried now. They are written for engineers, so each one has a plain summary here.
The essays
2 October 2026
When AI learns on the job, does the team learn too?
AI assistants can now keep what they learn from their own work and get better with it. What one person's assistant learns still doesn't reach a colleague's, and nobody seems to have built the thing that would carry it there. The essay looks at what the research has solved, what it hasn't, and what a fix would need.
Read the essay →2 October 2026
Who decides, and when?
The standards that let AI tools work together can say what a tool can do and who handles a task. They can't say whose judgement counts on a decision, or bring it in while the work is still cheap to change. The essay maps that gap and what would close it, and reports a test that didn't go the way we expected.
Read the essay →The notes
A few hundred words each, with a date and where the thinking stands.