Build an engine, not another programme
Build the capability to turn one AI success into the next — making each change faster, cheaper and easier to run yourself.
You don't need another pilot
Almost every large organisation is running AI pilots. Very few are compounding.
A pilot succeeds. It gets celebrated. Then it sits there, because nothing was built to turn one success into the next. The team disbands. The capability can't be reused. The next initiative starts where the last one did.
Organisations have been buying outcomes when what they need is the ability to produce outcomes repeatedly — because this time there's no end state to arrive at. That capacity is the engine.
Build an engine
An engine is the standing ability to redesign your organisation one service at a time, without commissioning a programme to do it.
It's four things you own: a queue of services worth converting, a method for converting one, shared capability every conversion reuses, and people ready to run what comes out.
A service enters. It comes out rebuilt with AI inside it — live, owned, measured — and its people already in new roles. Then the next one, faster.
Without one, every shift in the technology means another business case, another programme, another set of consultants.
Each discipline does a job in that cycle.
Take one away and it seizes. Redesign services without the workforce work and nobody's ready to run them. Do the workforce work without the backlog and you're reskilling toward roles that may never exist.
Why it compounds
The same steps repeat across every service. Built once, they're configured rather than rebuilt — the first conversion pays for them, the tenth uses them for nothing.
The same is true of the people work, and of the method itself. The first redesign is a discovery exercise. By the fifth, your teams are running it.
How it gets built
Foundations
Catalogue and backlog. Confidence across the workforce. The current deal assessed. One or two services converted end to end — enough to prove the method.
Conversion
Work the backlog. Stand up shared capability. Run transitions inside every conversion. Publish what changes when. Hand the method over as teams are ready.
Running
Services trigger their own redesigns. The capability library grows on evidence. The deal is reset on a rhythm.
Handover isn’t a phase at the end. The ability to keep adapting belongs inside the organisation from the start.
How you know it's working
- Conversions get faster and cheaper, not slower
- One improvement upgrades several services at once
- Services go back into redesign because their numbers say so
- People move into new roles, and the number is published
- Your teams run conversions without me
Why it matters
The organisations that pull ahead won't be the ones that adopted AI earliest. They'll be the ones that built the capacity to keep redesigning themselves while everyone else commissioned the next programme.