Writing
What we are learning.
What we are learning while we build an instrument for the human side of working with AI.
-
November 3, 2026
How many agents can you run at once before your judgment drops?
Everyone has a number of agents they run at once. Nobody measured it. What people say, what they do, and a check you can run on your own last week.
Read this -
November 2, 2026
Agents wrote more code. Firms shipped little more. Where did your review time go?
Harvard tracked 655 firms using AI coding agents. Commits jumped 36%, review wait times jumped 56%, and finished software barely moved. What did you approve?
Read this -
October 23, 2026
AI is removing the people who told you your work was good
Companies hire fewer beginners and give managers more people to look after. Recruiter Deepali Vyas explains why, and what it means for your own growth.
Read this -
October 22, 2026
Anthropic says your design judgment matters more than ever. OpenAI is teaching it to the model.
OpenAI grades GPT-6’s design judgment. Anthropic’s Meaghan Choi says yours matters more than ever. Four habits from her talk, and who grades yours.
Read this -
October 21, 2026
When Claude writes 90% of the code, what is left for the humans?
Anthropic shared how Claude Code ships 30,000 changes a day. Every fix makes the machines better. Five decisions are still yours, and how to practise them.
Read this -
October 20, 2026
Matt Pocock stopped writing code. The skill he says matters now is yours to practise.
Matt Pocock stopped writing code. The gist of his talk with Nate Herk, what he changed his mind about, and six practices to start this week.
Read this -
October 19, 2026
Theo says your prompts kinda suck. The fix is on your side of the keyboard.
Theo Browne says engineers overengineer their prompts. The gist of his new video, the examples worth knowing, and what it asks of the person.
Read this -
August 27, 2026
An AI Described How I Work. I Agreed. The Next Day I Threw It Out.
An AI described how I work and I agreed. A day later I threw it out. What a 1949 experiment and blood-pressure monitors say about every AI summary of you.
Read this -
August 24, 2026
A Passing Check Is a Claim, Not Evidence
Nine checks passed in three days of building with AI agents, and all nine were wrong. Why a green build is a claim, and where to look before you believe it.
Read this -
August 20, 2026
Breathing Is Not Your Strongest Habit
Sense’s first reading said my strongest habit was one everyone has, 40 of 40 times. Why counting tells you little, and what surgeons and pilots rate instead.
Read this -
August 17, 2026
How to Tell if You Are Steering the AI or Just Babysitting It
Give two developers the same feature on Monday. By lunch both are done, same tests, same pull request. Only one of them was flying the plane.
Read this -
August 14, 2026
The Day I Realized AI Was Changing My Behavior Without My Permission
Four hours in, my prompts had shrunk to "looks good, next". What that told us about working with AI, and why we built the feedback loop for ourselves first.
Read this