How we work
Evaluate. Improve. Confirm. We drive.
Three steps, in order, every time — and an operator of ours driving all three. The first is deliberately small and deliberately purchasable, because nobody should commit to a transformation program before anyone has established what is actually wrong.
- 01Scoped in a conversation
Evaluate
We start from what the business is trying to achieve, then assess your team’s talent, ways of working, tools, org structure, portfolio strategy, and where you genuinely sit on the five levels of AI capability — and tell you which of those is actually standing between you and the goal, whether or not you hire us to fix it.
- 02As long as the work takes
Improve
Our operators embed and do the work. Not a slide deck with recommendations — architecture, governance, hiring, migration, coaching, whatever the assessment says is required.
- 03Agreed up front
Confirm
On time, on budget, on content. We agree the measures before the work starts and report against them, so the improvement is a number rather than a feeling.
The fourth part
Someone has to drive.
Three steps do not execute themselves. The gap between a plan and a result is almost never the plan.
Most organizations we meet do not lack analysis. They have the deck, they broadly agree with it, and eighteen months later the same constraint is still there — because the work required a decision nobody owned, a conversation nobody scheduled, and a person willing to keep asking for both.
That person is what we supply. Our operators have held the roles they are working in, so they know what it takes to get something done inside an organization of this size: which decision is actually blocking, who has to be in the room, what has to be true before the next thing can start, and when a plan needs to change because the ground did.
It is the difference between a firm that reports on progress and one that is accountable for it. We take the thing and move it, and we keep moving it until the measure agreed in step three says it moved.
Why it is in this order
Diagnosis is a deliverable, not a sales call.
Most consulting engagements begin with a proposal for work that was scoped before anyone looked. That is convenient for the firm writing it and expensive for the organization paying for it, because the visible symptom is very rarely the cause.
So we sell the diagnosis separately. A Jump Start has a scope and a length agreed with you before it starts, and produces a written assessment that is yours whether or not you engage us for anything else. If the answer is that you do not need us, that is a legitimate outcome, and we would rather you found it out at the start than two quarters in.
An assessment is only useful if it also says what to do first, and nothing in the evidence decides that on its own. So the order comes from you: we agree at the outset what the business is trying to achieve — the launch you have committed to, the cost you have to take out, the audit you have to pass — and we rank the constraints by what each one is costing that, not by which is easiest to fix or most interesting to us. Two organizations with the same findings get different sequences, and both can check the order against something they said out loud.
What follows is not a recommendations deck. Our operators embed and do the work — architecture, governance, hiring, migration, coaching, whatever the assessment said was required, in the order the goals put it in. Then we report against measures agreed before any of it started, so the improvement is a number rather than a feeling.
AI, in the same three steps
No separate AI methodology.
The reasons AI programs stall are, item for item, the things we already fix. So the AI work sits inside the model rather than beside it — which is the strongest argument that it belongs.
Evaluate
Which level are you actually at, versus the level your roadmap assumes? Is the data good enough to support the next one? Does the organization change fast enough to absorb it?
Improve
Build the layer that was skipped. Usually data quality, delivery discipline, or governance long before it is a model — which is precisely why the firms selling models alone keep failing.
Confirm
Eval-driven measurement. Outcomes, not activity metrics. Did the AI spend return anything, measured against targets agreed before the work started?
Two hours to find out whether we recognise your problem.
No deck, no obligation. We listen, we tell you whether we have seen this before, and we say what we think it would take. If a Jump Start is the right next step we will say so — and if it is not, we will say that too.