Evaluate
AI readiness assessment
A 360-degree view of whether your organisation can actually absorb AI — across people, process and technology, with governance and measurability running through all three. It ends with a readiness position per dimension and the specific gaps blocking your first use case.
Duration
4–5 weeks
Built on
ISO/IEC 42001 · EU AI Act · NIST AI RMF
Indicative price
€8,500–19,000 per engagement
Who this is for
CEO or owner
Is being told the company must do something about AI and wants to know what is real.
COO
Will own the operating change and needs to know whether the organisation can take it.
CIO or CTO
Knows the data situation is worse than the enthusiasm suggests.
HR or transformation lead
Has to build the capability and needs a baseline to build from.
Readiness is not an appetite question
Almost every organisation reports being ready for AI. What they are describing is appetite — the board is interested, someone has run a pilot, the enthusiasm is real. Appetite is not readiness. Readiness is whether the organisation can absorb the change: whether the data exists where people think it does, whether the process being automated is stable enough to automate, whether the people expected to work differently have the literacy to, and whether anyone owns the result once the project ends.
Pilots that fail almost always fail for reasons that were knowable beforehand. The data turned out to live in three systems with different definitions of a customer. The process being automated was never documented and varied by whoever was doing it. The model worked and nobody owned it after the consultant left. None of these are AI problems and all of them are discoverable in four weeks.
So we assess three dimensions rather than one. People: literacy by role, capacity to absorb change, and the posture of the middle management layer who will quietly decide whether any of this is used. Process: whether candidate processes are documented, stable and measured — because an AI applied to an undocumented process does not automate the process, it automates one person's habits. Technology: whether the data is available, of usable quality, accessible under the right permissions, and traceable to a source; and whether there is anywhere appropriate to run the workload.
Two things cut across all three. Governance: who decides which use cases proceed, who is accountable when one misbehaves, and whether there is a policy that describes the company you actually are. And measurability: whether a baseline exists for the thing you intend to improve, because a use case with no before is a use case with no after.
The output is a readiness position per dimension and, more usefully, the specific things blocking your first intended use case. Sometimes the honest finding is that a data problem has to be solved before any AI is worth attempting. That is an uncomfortable conclusion in week four and a very cheap one compared with month nine.
How we do it
- 01
Framing and scope
3 days
What the organisation is actually trying to achieve, which functions are in scope, and whether there is a candidate use case already in mind. If there is, the whole assessment is pointed at it; if not, choosing one is part of the work.
- 02
People
1 week
AI literacy baseline by role, capacity interviews, and an honest read of the middle management posture. We also inventory what is already being used without sanction — this is usually the most informative hour of the engagement and it is never in anyone's records.
- 03
Process
1 week
The candidate processes: are they documented, how much do they vary between people, are they measured today, and is the variation a defect or the actual value of the human doing them.
- 04
Technology and data
1.5 weeks
Data availability, quality, lineage and access rights. Integration surface. Infrastructure, and where a workload could legitimately run given your obligations. We assess whether the data is usable; we do not need the data itself.
- 05
Governance and measurability
3 days
Decision rights, accountability, existing policy, indicative EU AI Act exposure, and whether baselines exist for what you intend to improve.
- 06
Scoring and readout
3 days
Readiness per dimension, the blockers named specifically, a sequenced plan, and an executive readout that says plainly whether to proceed now, proceed narrowly, or fix something first.
Named artefacts
What you receive
- Readiness scorecard across people, process, technology, governance and measurability
- Shadow AI inventory — what is already in use, by whom, on what data
- AI literacy baseline by role, usable as a before-measurement later
- Process candidate assessment — documented, stable, measured, and the variance in each
- Data readiness assessment — availability, quality, lineage and access rights
- Infrastructure and workload placement assessment
- Governance gap summary with indicative EU AI Act exposure
- Blocking issues for your first intended use case, named individually
- Sequenced readiness plan with owners and effort
- Executive readout with a clear proceed, proceed narrowly, or fix first
What we need from you
- Eight to twelve people for an hour each, across functions — not only IT. The finance and operations conversations are usually the revealing ones.
- Honesty about what is already being used without sanction. We are not there to police it, and an accurate inventory is worth more than a policy.
- Access to data documentation, or an acceptance that discovering there is none is itself a finding.
- Your candidate use case if you have one. If you do not, we will help you choose, but the assessment is sharper when pointed at something real.
- An executive prepared to hear 'not yet' and act on it.
What changes
- 01You know whether you are ready, per dimension, rather than as a general impression.
- 02You know what specifically blocks your first use case, rather than that it is 'complicated'.
- 03Shadow AI is an inventory rather than a rumour.
- 04You have a literacy baseline, so the training you buy later can be shown to have worked.
- 05If the answer is not yet, you have it in week four instead of after a failed pilot.
What it costs
€8,500–19,000 per engagement
All prices exclude VAT.
Questions
How is this different from the AI value assessment?
This one asks whether you can. The value assessment asks what it is worth. Most organisations need both, and we usually run readiness first — a value map you cannot execute is an expensive document. If budget allows only one, take this one.
What if we score badly?
That is a useful outcome, not a failed engagement. The two most common blockers are data access and middle-management capacity, and both are fixable in months rather than years. The expensive path is finding them after committing to a pilot and a vendor.
Do you need access to our actual data?
No. We assess whether data is available, documented, of usable quality and accessible under the right permissions. That can be established from documentation, system metadata and interviews without us handling the data itself.
Is this the same as EU AI Act readiness?
No. That establishes your legal obligations and your role under the regulation. This establishes organisational capability. They answer to different people and are sold separately, though this one flags indicative exposure so you know whether you need the other.
How long before it is out of date?
The technology and data findings hold for a year or so. The people findings move faster, which is why the literacy baseline is worth keeping — it is the thing you measure against after the training.
Leave with your top three risks documented
Thirty minutes with a senior practitioner. No slideware, no sales engineer.