An AI strategy is useful only if it says what to do first, what it costs, what has to change in the data and systems, and who is accountable. Most strategy documents stop at ambition.
We assess where you stand on data, platforms, governance and skills, and produce a costed roadmap of use cases you can actually deliver.
Use cases in order, with cost, dependencies and owners.
Data, platform, security, governance, skills and operating model.
Acceptable use, risk classification and approval process.
Your data is not used to train public models, and access follows your permissions.
Six workstreams, run together, because the gaps in one usually explain the stalls in another.
Candidate use cases gathered across the business, scored on value, feasibility, risk and running cost, and sequenced.
Quick wins and foundational work balanced in the order.
A portfolio, not a wish list.
Data sources mapped for each priority use case, with quality, ownership, access and freshness assessed.
The most common blocker, and the one least visible in a pilot.
Data gaps known before they cost money.
Target AI architecture across model providers, cloud platforms, vector search, integration and the copilots in software you already own.
Designed to avoid lock-in where it matters.
One platform pattern for every use case.
AI acceptable use policy, risk classification, approval process, model and vendor assessment, and alignment with data protection law.
Proportionate: low-risk uses move fast, high-risk uses get review.
Rules people can follow.
Who owns AI, who builds, who approves and who runs it, with training for leaders, builders and users.
A centre of enablement rather than a bottleneck.
Clear ownership from idea to operation.
A twelve to eighteen month plan with costs, dependencies, milestones and measures of success.
Written to go to the board as it stands.
A plan you can fund and track.
How a strategy engagement runs
The readiness assessment measures six dimensions against evidence.
| Dimension | Question | Evidence reviewed | Output |
|---|---|---|---|
| Data | Is the data usable for AI? | Sources, quality, ownership, access | Data gap list |
| Platform | Where will AI run? | Cloud, licences, integration layer | Target architecture |
| Security | Can it be used safely? | Identity, data classification, logging | Control gaps |
| Governance | Who decides? | Policies, approvals, risk process | Governance model |
| Skills | Who can build and run it? | Teams, partners, training | Capability plan |
| Operating model | Who owns it in production? | Support, budgets, measures | Ownership model |
Strategy documents fail when they describe ambition without the dependencies: the data that is not ready, the platform not chosen, the policy not written, the owner not named.
Ours ends with a costed roadmap and a scoped first project, so the strategy turns into delivery while the people who wrote it are still in the room.
Plan an AI strategyScope depends on the number of business units and systems involved.
Not always. A low-risk use case with good data can start now. A strategy matters once several teams are buying or building AI independently, or when the board wants a plan.
Yes. An acceptable use policy, risk classification and approval process are part of the governance workstream.
Yes. Copilots and AI features in Microsoft, Oracle and Salesforce are often the fastest route to value, and they are assessed alongside custom builds.
A sponsor, process owners from the priority areas, and representatives from IT, security, data and legal.
You can deliver it with your own team, with us, or with others. The roadmap is written so any capable team can execute it.
Tell us where AI is being discussed in your business today. We will tell you what an assessment would cover.