Services · 4–6 weeks
Separate the AI work worth funding from the work that only demos well.
A fixed-scope engagement, four to six weeks, that establishes what your data can actually support, which decisions are currently being made without evidence, and what to build in what order.
The hard part is not choosing a model. It is choosing the problem.
Most organisations arrive at AI with a shortlist of candidate use cases assembled from vendor conversations and internal enthusiasm. The list is usually plausible and rarely ranked by anything meaningful. What is missing is an honest read on which items are blocked by data that is not reachable, which are blocked by a process nobody wants to change, and which are genuinely close.
Assessments that skip this produce a strategy deck and a pilot. The pilot demonstrates that the technology works, which was never seriously in doubt, and then stalls at the point where it would need real data from a real system. The programme loses eighteen months and, more damagingly, its internal credibility.
The assessment exists to fail candidate use cases cheaply and early, so the ones that survive are worth the build.
What we build
What the four to six weeks contain
Evidence inventory
Where the organisation’s decision-relevant data actually lives, in what state, behind what access model, and how reliably it is maintained. This is done by connecting to the systems, not by asking people to describe them.
Integration feasibility
A concrete read on each target system: is there an API, is it enabled by policy, what does authentication require, and what is the realistic effort. Systems that look accessible on paper are frequently not, and it is much cheaper to learn that now.
Decision mapping
The recurring decisions your teams make, what evidence they currently rest on, and what it costs when they are wrong. AI value is easiest to defend where it improves a decision somebody is already accountable for.
Candidate evaluation
Each use case scored on data readiness, integration cost, decision value and governance exposure — with the ones that fail marked as failed and the reason recorded, so they are not rediscovered next year.
A working proof on real data
Where the assessment finds a strong candidate, we build a narrow proof against your actual data rather than a demonstration environment. A working artefact settles arguments that a document cannot.
Costed roadmap
A sequenced plan with effort estimates, dependencies, the shared substrate each phase contributes to, and an honest note of what we would not build. Yours to execute with us, with someone else, or internally.
What changes
What you have at the end
A defensible answer for your executive sponsor
Not a maturity score. A specific list of what to build, what it costs, what it depends on, and what you will be able to do afterwards that you cannot do now.
Integration risk retired early
The systems that would have derailed month seven are identified in week two, when the plan can still change cheaply and without anyone losing face.
A shorter list, held with more confidence
Fewer candidate use cases than you started with, each surviving a real test rather than an enthusiasm test.
No obligation to continue
The deliverable is complete and portable. If the right next step is your own team, that is a legitimate outcome and we will say so.
What we need from you
Read access to the relevant systems, a few hours with the people who make the decisions in scope, and one executive sponsor willing to hear that a favoured use case is not ready. The last of those matters most.
Start with a conversation about your environment.
Thirty minutes is usually enough to tell whether an assessment is the right first step or whether you already know what to build.