Perspectives
What we're seeing in governed AI adoption.
Essays on where enterprise AI actually stands: what the data shows, what boards should require, and how we structure engagements against it. Every statistic carries its source.
- 2 July 20263 min read
The deployment is the product
Capability keeps rising and agentic programmes keep getting cancelled. After twenty-five years of building enterprise systems, I recognise the pattern — organisations keep buying the agent when what succeeds or fails is the deployment.
- 2 July 20263 min read
The token party is over
At comparable capability, completing the same task can cost twenty times more on one model than another. I treat model choice as a financial control, and I think your CFO should too.
- 2 July 20264 min read
Value shows up where it is measured
88% of organisations use AI; 5.5% report real financial returns. I read that as a measurement gap before a capability gap, and I have an interest in saying so — which is exactly why the method has to be one you can hold me to.
- 2 July 20263 min read
Your model is a supplier
Boards manage concentration risk in every material supplier except the most concentrated one in their AI programme, the model. I spent years inside institutions that drilled supplier failure; the hedge here is architectural, and it is cheap only if you adopt it early.
Begin
Book a scoping conversation.
A free working conversation on where you stand, and the first move that fits.