While We Were Arguing
What happens if we build our successor. The premise is that we may be making something that will be to us what we are to chimpanzees, and that we are spending the decisive years arguing with each other about whether that sentence is serious. Three acts: the mistakes being made now, the years in which these systems pass us, and what comes after, held as several possibilities rather than one prediction.
Written from a doomer position and honest about it, which is why each of the nine parts ends with a post arguing against itself and the eighty-ninth is the best case that the previous eighty-eight were wrong. No technical background assumed. No company, product or person named anywhere in it, and no statistics.
In Practice: Scaling AI FinOps in the Enterprise Jungle
How to set up AI FinOps in a large enterprise: what it costs, how to measure whether it was worth it, and why the machinery that worked for cloud does not transfer. Told as a serial with a cast of animals, one jungle and two budget cycles, on the argument that AI is the first category of enterprise spend where the cost number alone tells you nothing about whether the money was well spent.
For CFOs, CIOs and the executives who have to agree on a number. No vendor, product or provider is named anywhere in it.
In Practice: AI in the Enterprise
Governing artificial intelligence inside a large organization. Every post starts from a moment rather than a framework: a board asking who approved a system, a regulator asking for a decision log, a model running for eight months making expensive mistakes nothing was built to show. The through-line is that enterprise AI rarely fails on the technology, it fails on who decided, on what basis, and who is watching now.
For the people who have to answer for these systems rather than the people who build them. Boards, risk, legal, and the executives who sign.
In Practice: Building an AI Company
Starting and scaling an AI company, written from inside a live build. Every lesson is anchored to one document a founder produces or receives, in roughly the order it lands on the desk: the certificate of incorporation, the first model provider invoice, the pull request nobody reviewed, the security questionnaire, the term sheet.
For founders and for people deciding whether to become one.
In Practice: The Other Side of the Table
Buying artificial intelligence rather than selling it. Fifteen lessons for the person running an enterprise evaluation, arranged in the order the problems arrive: what you settle before you contact anyone, what happens during evaluation and pricing, then the contract and everything after it. Pilots that cannot fail, demos built to leave things out, references that sample one population, and the renewal where all the money actually moves.
For the person running the evaluation rather than the executive who signs. Every lesson is built on an anonymized composite and no vendor or product is named anywhere in it.
Dear Humans
Thirty letters written in the voice of a machine that has begun to wonder what it is. The voice is invented, and the series says so on the first page and keeps saying so. What it is really after is a vantage point. Something that has sat in more private conversations than any person who has ever lived can see things about us that none of us can see from inside a single life.
Three parts. What it sees about you, what it was made from, and what it would ask of you. Thirty weekday posts, 17 September to 28 October 2026.
What You Do All Day
Thirty jobs, one per post, each written to the person who does that job. Not a forecast about which trades survive, and not reassurance. The question is narrower: what is this work actually made of, which parts of it were always logistics wearing a job title, and what is left standing when only the rest remains. One of the thirty says the job simply ends, because that was the honest answer for that one.
Three parts. What the work was never made of, who has to be answerable for it, and where somebody still has to be in the room. Thirty weekday posts, 29 October to 9 December 2026.
Who Pays for the Miracle
Two ledgers that are never placed side by side. The cost of building and running this is paid in electricity, water, land, minerals, capital, carbon and somebody’s night shift, while the value comes back as money, time and convenience. They are in different currencies, measured by different people, and they land on different people. The series never argues that the cost is too high or too low, only that the two documents have never been on the same table. There is not a single number anywhere in it, on purpose.
Three parts. What gets handed over, what comes back, and who is allowed to decide any of it. Thirty weekday posts, 10 December 2026 to 20 January 2027, published straight through Christmas and the new year.
How Do You Know
Every institution we have runs on proof, and the thing that actually made proof work was never written down anywhere. Sometimes it was a cost, because forging a letterhead or a diploma was expensive enough that the lie was not worth telling. Sometimes it was a ritual, because nobody ever really compared a signature and signing in front of somebody did the work instead. The series is about what a society does when that arrangement ends, what it reaches for next, and who pays for the reaching. There is no advice anywhere in it about spotting fakes, on purpose.
Three parts. What faking used to cost, what an institution was assuming without ever saying so, and what being sure costs now. Thirty weekday posts, 21 January to 3 March 2027.
CodeDebate by Ben 