How technology changes the work, and what it takes to build something useful.
The PM who understands AI deeply isn't just more efficient than one who doesn't. They're doing a categorically different job. Most product functions haven't caught up with what that means yet.
Archive
Most AI transformation stories are told at the end, when the numbers look good. This is the version that includes what day one actually looked like, what broke first, and what it took to get from chaos to a system that held.
The single biggest unlock in how I work with AI day to day wasn't a smarter model. It was building one memory layer that every tool I use, including my own notes, reads from and writes to.
Every improvement in AI capability makes the next layer of human judgment more, not less, important. Understanding which layer that is might be the most consequential strategic question of the next decade.
The data on CEO AI decision-making in 2026 is striking. Not because CEOs are taking AI seriously, they are. But because the gap between the weight of the decisions and the quality of the support available to make them is growing, not shrinking.
The CAIO deadline is weeks away. Most appointees have the title and the accountability. The practical question of what to actually do on day one is less well answered.
The public mental model of AI development is still one person, one chatbot, one output. The reality is a coordinated pipeline of specialist tools, each doing the thing it's genuinely best at.
The most important AI conversation your board will have isn't about capability. It's about whether the organisation is actually ready to benefit from it, and most boards are getting a sanitised version of that answer.
Most infrastructure discussions start with what the stack can do at scale. The more interesting question for anyone building an AI product is what it costs before you get there.
The org chart is being redrawn. Not by cutting people and leaving the gaps, but by replacing layers of execution with AI agents and elevating a single person to run the whole operation.
In a metro market, early AI adoption gives you an edge over competitors who are also adopting AI. In a regional market, it gives you an edge over competitors who aren't thinking about it at all.
The build-vs-buy decision in AI is more consequential than it's ever been, and most of the advice on it comes from people trying to sell you one of the two options.
Notes on building with AI, product decisions, and lessons from practical work.
No spam. One useful note when there is something worth sending.