Scored use-case inventory
Every candidate, its scores, its owner and the evidence behind the numbers. The single most reused document we produce.
AI Services · Strategy & leadership
AI strategy that ends in a funded first project
We score your use cases on value and feasibility, check whether the data can support them, and hand you a 12-month roadmap with a business case your board can approve.
- 4 to 6 weeks
- Fixed price
- Vendor-neutral roadmap
Who this is for
Too many ideas, no agreed priority
Leadership teams come to us with the same three problems. There are a dozen AI ideas and no agreed way to rank them. Vendors are promising outcomes nobody can verify. And the finance team wants a number before anyone has defined the work.
Our strategy engagement fixes the ranking problem first, because everything else follows from it. Each use case is scored on the same criteria, in the open, with the people who will own it in the room. The result is a short list everyone can defend and a long list everyone can live with.
- A board or executive team under pressure to "do something with AI"
- A CTO or CIO who needs an independent view before committing budget
- A product leader with customer demand for AI features and no architecture
- A transformation lead with a pilot graveyard and no measurement
The scoring method
Six criteria, scored in the open
Each use case is scored from one to five on six criteria by the people who would own it, with us challenging the evidence. The totals rank the list; the pattern of scores explains why.
| Criterion (1–5) | Example: support deflection | Example: invoice extraction | Example: demand forecasting | Example: autonomous negotiation |
|---|---|---|---|---|
| Business value (cost saved or revenue) | 5 | 4 | 5 | 3 |
| Technical feasibility today | 4 | 5 | 2 | 1 |
| Data readiness (exists, clean, accessible) | 4 | 4 | 2 | 2 |
| Risk and compliance exposure (5 = low) | 4 | 4 | 4 | 1 |
| Time to first measurable result (5 = fast) | 4 | 5 | 2 | 1 |
| Organisational readiness (owner, process, appetite) | 4 | 4 | 3 | 2 |
| Total (max 30) | 25 | 26 | 18 | 10 |
| Quadrant | Do now | Do now | Plan and de-risk | Avoid |
Illustrative scores for example use cases. Real engagements score 10 to 30 candidates and weight criteria to the client's situation. Regulated industries weight risk more heavily; see our industry pages.
What we deliver
Six documents, one decision
Target architecture
Model policy, retrieval approach, integration points and hosting choices, drawn as the system you will actually run. See LLM app development.
Data readiness findings
Where the data for the top use cases lives, who owns it, how clean it is and what it will take to make it usable. Read our guide to data readiness for AI.
12-month roadmap
Sequenced projects with budget ranges, dependencies and decision points. Written so the second project can start before the first is perfect.
Business case for project one
Costs, expected benefit range, the metric that proves it and the point at which you would stop. Built with our approach to measuring AI ROI.
Governance baseline
An AI policy outline, a risk register and the controls you will need for the EU AI Act or ISO 42001, via our AI governance practice.
Process
Four to six weeks, four stages
Discover (week 1–2)
Interviews with leaders, process owners and engineers. Inventory of systems, data and existing pilots. Candidate use cases collected, not judged.
Score (week 2–3)
Scoring workshops with owners. Data inspection for the top candidates. Build-versus-buy checks against the market.
Design (week 3–5)
Target architecture, model policy, security and governance baseline. Cost ranges for the top three projects.
Roadmap (week 5–6)
Sequenced plan, business case for project one, leadership readout and a decision meeting with a yes-or-no question.
What goes wrong when strategy is skipped
The most common failure is starting with the most exciting use case rather than the most feasible. Six months and a large invoice later, the team discovers the data was never clean enough, the integration needs a system nobody owns, or the compliance team was never asked. Scoring in week one costs a fraction of that and makes the same discovery on paper.
How to decide
Build, buy or wait?
Every use case on the roadmap carries a build, buy or wait recommendation. Buy when a vendor product covers most of the job and your data is not the differentiator. Build when the workflow is specific to you, the data is yours and the economics at volume favour owning it. Wait when the value depends on a model capability that is improving quickly, or on data you do not yet have. Our guide to build versus buy for AI works through the trade-offs in detail.
Fractional AI CTO
Keep the strategy honest month after month with a part-time technical leader.
ExploreAI Implementation
Deliver project one with the team that wrote the roadmap.
ExploreAI project roadmap guide
How we sequence projects and set decision points.
ReadEngagement models
Fixed scope, retainer or team extension: how the commercial side works.
Compare
FAQ
Frequently asked questions
How long does an AI strategy engagement take?
Four to six weeks for a single business unit, eight to ten for a group with several divisions. The limiting factor is usually access to the people who know where the data lives, so we book those interviews in week one.
What do we get at the end?
A scored use-case inventory, a target architecture, data readiness findings, a 12-month roadmap with budget ranges, a business case for the first project and a governance baseline. All written to be read by a board and used by an engineering team.
Do we have to build with you afterwards?
No. The roadmap is vendor-neutral and you can take it to any team. Many clients do ask us to deliver the first project through AI implementation, which is why we only recommend things we would be willing to build.
We already have a list of AI ideas. Is this still useful?
Usually yes. The list is rarely the problem; agreeing which three to fund and which seven to park is. The scoring method gives leadership a defensible way to say no, and the data readiness check stops the favourite idea failing on week two.
How is this different from the free AI readiness assessment?
The readiness assessment is a ten-minute self-check that gives you a score and a shortlist of gaps. Strategy consulting is a facilitated engagement with interviews, data inspection, architecture work and a funded plan.