Frameworks
The named ideas.
Naming a pattern is not a rhetorical flourish. A pattern with a name can be pointed at in a board meeting, argued with, and recognised the second time it appears. These are the ones I keep returning to, each with a permanent home here.
The Pilot Trap
The most dangerous moment is not when the pilot fails. It is when it succeeds. A successful pilot generates the political capital to keep piloting, and organisations mistake that motion for progress. The trap is that the conditions which make a pilot succeed — a hand-picked team, a clean dataset, an absence of integration constraints — are precisely the conditions production will not provide.
The Technology-First Trap
Starting from the capability rather than the constraint. It looks like rigour — evaluations, benchmarks, architecture diagrams — and it produces a solution in search of the problem that justifies it. The tell is an organisation that can describe its stack in detail and its bottleneck only vaguely.
The Solo Trap
The belief that a transformation of this scale can be run inside one organisation’s boundary. It is the most expensive of the three, because it is invisible until the integration work arrives and there is no ecosystem to absorb it.
Different verticals. Same wall.
Manufacturing, automotive, healthcare, retail and banking describe their challenges as though they were industry-specific. They are not. Each arrives at the same orchestration wall from a different direction, which is why the vertical framing obscures more than it reveals — and why the lessons transfer far better than most leaders expect.
The Workload Sovereignty Matrix
Sovereignty is usually debated as a binary and a compliance problem. It is neither. It is a per-workload decision with distinct axes — data gravity, regulatory exposure, latency tolerance, and the cost of being wrong — and the matrix exists to stop organisations answering it once, globally, and living with that answer for a decade. Developed in The Token Trap.
The Intelligence Control Stack
What an organisation actually controls when it deploys intelligence it did not build. Working through the layers separates the parts that are genuinely yours from the parts you are renting under someone else’s terms — a distinction most enterprise AI strategies never make explicit. Developed in The Token Trap.
Each of these will expand into its own page, with a contrast diagram and the field evidence behind it.