Insights, reflections and experience from the practice of interim project and programme management.
Anthropic's biggest partners are at once its customers, suppliers, shareholders and competitors, and its commitments run far ahead of its revenue. I looked at who carries the risk if growth slows, what all that compute is actually used for, and why I think the cost of making AI reliable inside organisations hasn't been priced in yet.
Read more →Anthropic's prospectus shows real growth, but it doesn't show how its customers will earn back what they spend. A look at the prospectus through two earlier waves I lived through in banking, offshoring and Agile, and at why the hardest part of AI adoption may still be ahead.
Read more →Most mornings I open something with a good title and know within ten lines that there is nothing under it. This is about why that happens, why I do it too, and what writing is actually for. Written with a language model, which is part of the argument.
Read more →Last month I traded crash predictions for a watchlist. Four weeks later, Alphabet posted its first negative free cash flow quarter since 2004 and Anthropic reported a 65 billion dollar run rate, so I ran the review early. What moved, what refused to move, and why the only numbers that would settle the debate still don't exist.
Read more →My timeline keeps predicting the end of the AI party, so I checked the doom claims against the actual numbers. They get more right than I expected - but almost nobody says what would prove them wrong. So instead of a prediction, I made a watchlist.
Read more →Sander Schimmelpenninck always writes sharply about divides and inequality, until he writes about AI, and then he misses the way out. My point is simple: AI does not compel, it polarises. In my view the question is not whether we can refuse the technology, but how we teach people to steer it.
Read more →English language article about the way frontier models like the new Fable relate to their older cousins and the local LLM capabilities you might want to be using instead.
Read more →A long read on the temptation of hosting AI locally, the sum that seems to add up, and the question that should lie beneath it.
Read more →How I built a team of eleven AI specialists in five days that operates autonomously every day, without writing any code myself.
Read more →Following my article of last week on the challenges of using AI in the heavily regulated banking sector, an excursion to a profession that has traditionally been less accustomed to seeing IT as part of its core activity: the legal profession. Strict rules apply here too, and supervisors are vigilant about improper or incorrect use of AI.
Read more →Dutch custodian banks use AI for monitoring, onboarding and trading, but the conversation about governance and the AI Act has barely started. An analysis of what the Investment Officer article did not say.
Read more →My colleague discovered how vibe coding works and went all in: seven projects in two evenings, all at once, none finished. The buckshot method has something tempting and rational about it, but a new trap sits inside it.
Read more →Among people who use AI tools, a divide is emerging that has little to do with technology. One group knows what these tools make possible, the other does not (yet). On the gap between building and watching.
Read more →Two people use the same AI tool on the same hosting environment. One deploys without problems, the other gets stuck and concludes that the tooling does not support it. The one who got stuck had more technical knowledge. I call this the expertise paradox of vibe coding.
Read more →Boris Cherny, development manager of Claude Code at Anthropic, has not written a single line of code since November 2025. I am a programme manager, not a developer, and I have still built two working web applications. Below I describe how that was possible and what I think it means for the market.
Read more →As a senior professional you are increasingly forced to work through a payroll company or secondment agency, and none of the usual options feels like a step forward. What if the intermediary adds value? On the Inductus model as an alternative to opaque payroll constructions.
Read more →