Writing

I write about getting AI to work inside a company: the systems it has to use, the people responsible for its mistakes, and what happens after the demo. The personal essays follow questions from the rest of my life, including attention, games, films, and how I make sense of things.

Every published piece. Select a column heading to sort by it; select again to reverse.
Opening image
The Months After Go-Live Following adoption long enough to see whether the work changes change management Helping people adopt a system after delivery
Working Out What an Agent Will Cost The estimate changes with workload, review time, and where each task runs model economics An AI bill that outruns the business case
Everything I Cannot Unsee What twenty years of caring about details did to how I watch a film expertise Noticing every flaw and having nowhere to put it
Migration Archaeology Finding the decisions and workarounds a replacement needs to account for legacy systems Understanding an established system before replacing it
Two Monitors, One Mind How I use the second screen, and what I give up when the television is on attention Choosing what to put on a second screen
Before an Agent Reads the Customer Records Finding the corrections people make before trusting the data master data Checking the records an agent will rely on
Before I Let an Agent Do More How I would examine the tests behind a proposal to expand its authority evaluation Evaluating an agent before expanding its authority
The R-Tune Button My brother and I used to find it easier to sit through a film attention Finding it harder to settle into a film
Bringing an Agent Proposal to Finance The questions I would prepare for before asking someone to sign off finance systems Preparing an agent proposal for financial review
What Manufacturing Taught Me About Agent Failure Modes The shop floor has been running autonomous systems against physical reality for decades manufacturing Agents that fail quietly instead of stopping
The Reversibility Ladder Where agents may write in a system of record governance Deciding what an agent may safely write
Seven Writes I Would Keep Outside an Agent's Authority The permissions I would exclude at the beginning of an ERP project controls Setting agent guardrails before anyone builds
The Stack and the Spell How a card game became part of the way I learn analogy Using ideas you cannot fully explain
Getting an AI Pilot Into Production The integration work I would include in the first estimate integration architecture Taking a useful pilot into production
The Bonfire Problem What it was like to feel other people's emotions before I had a word for it sensitivity Feeling other people's emotions more than you can carry