Service 03
Strategy, platform, retrieval and governance across departments, with each stage proving its governance before the next begins. This is how agent sprawl is prevented rather than cleaned up afterwards. Each stage leaves a governed system and a baseline the next stage can be measured against.
The Stages
Each stage is separately priced and independently valuable. Nothing requires a commitment to the next one.
Stage 0
Where AI genuinely fits and where it does not. Use cases scored by feasibility, consequence and cost, then sequenced by what delivers value soonest. Output: a roadmap with honest ceilings per department, and the ones we recommend against.
Stage 1
The first system, in the department where the knowledge problem is sharpest. Retrieval that cites its sources and refuses when it cannot answer. This is deliberately first: it is the lowest-risk system that produces visible value.
Stage 2
One department to AI-native operation against its stated ceiling. Governance active from the first deployment. This stage is where the organisation learns what the method actually costs and delivers.
Stage 3
The next department, with the governance model already proven and the internal team now able to run parts of it. Each subsequent department costs less than the one before.
Why Sequencing
Many AI projects, several teams, and no single layer that governs any of it. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027, citing rising costs, unclear business value and inadequate risk controls (source).
Problem 01
Systems are built, then someone asks who governs them. Adding provenance and authority limits afterwards means rebuilding, which nobody funds. So it does not happen.
Sequenced: governance ships with the first system, and the model is reused.
Problem 02
Six teams solve the same retrieval problem six different ways, because they ran in parallel and none of them finished first.
Sequenced: department two inherits what department one proved.
Problem 03
Everything started at once, so there is no before-state to measure against. Eighteen months later nobody can say what the spend produced.
Sequenced: each stage has a baseline, because the previous stage established it.
One at a time
Not two. This is a limit on our capacity, stated plainly rather than discovered by a client in month four.
Two weeks, fixed price. You get the roadmap, the honest ceilings per department, and the list of things we recommend against doing at all.