Evaluate
AI value assessment
Where AI would create value in your business, what each opportunity is worth on the create–deliver–capture chain, and what it would cost to build, run and own. A scored and sequenced portfolio you can fund — not a list of ideas.
Duration
3–4 weeks
Built on
ISO/IEC 42001 · NIST AI RMF
Indicative price
€8,500–19,000 per engagement
Who this is for
CEO or owner
Needs to know what AI is worth here before funding anything.
CFO
Wants a business case with a number in it.
COO
Knows where the work piles up and wants to know which of it is addressable.
Head of a function
Has a specific problem and wants to know whether AI is the right instrument.
Everything looks like a use case until you price it
Ask any organisation for AI opportunities and you will get thirty. Ask which are worth doing and the conversation stops, because ranking them requires three things most lists do not carry: what the opportunity is actually worth, what it would cost to build and run, and how likely it is to work given what you have.
Without that, prioritisation defaults to sponsorship. The use case with the loudest advocate goes first, which is unrelated to the use case with the best return, and the programme's credibility is spent on whatever that first one produces.
The scoring has to be honest about cost. A use case is not the model; it is the data preparation, the integration, the change management, the monitoring, and the person who owns it afterwards. Most published estimates of AI cost describe only the cheapest part.
It also has to be honest about value, and this is where discipline matters most. Value created is not value captured. An efficiency gain that frees four hours a week is worth nothing unless those hours go somewhere that matters, and a model that improves a forecast is worth nothing if nobody changes a decision because of it. We score on the full chain — create, deliver, capture — and a use case that breaks at the last link is marked accordingly.
The output is a scored portfolio, a recommended first move with the reasoning, and a clear statement of what each one needs to be true before it starts.
How we do it
- 01
Opportunity discovery
1 week
Structured sessions by function, plus a look at where time and error actually concentrate. We collect more than we will keep.
- 02
Qualification
3–5 days
Each candidate tested for whether the data exists, the process is stable enough, and the outcome is measurable. Most of the thirty stop here, and that is the point.
- 03
Value modelling
1 week
What each survivor is worth on the create–deliver–capture chain, with the assumptions written down so they can be argued with.
- 04
Cost and feasibility
3–5 days
Build, integrate, run and own. Including the monitoring and the person accountable after go-live, which is where most estimates quietly stop.
- 05
Scoring and sequence
3 days
Ranked on value, feasibility and regulatory exposure, with the dependencies between them made visible.
- 06
Readout
half a day
The portfolio, the recommended first move, and what has to be true before it starts.
Named artefacts
What you receive
- Opportunity long list, with the reasons for every rejection recorded
- Qualified use case portfolio, scored on value, feasibility and regulatory exposure
- Value model per use case, with assumptions stated
- Cost model covering build, integration, run and ownership
- Dependency map between use cases
- Recommended first move with the argument for it
- Preconditions per use case — what must be true before it starts
- Executive readout pack
What we need from you
- Function leads for the discovery sessions. The opportunities are in their heads, not in a system.
- Financial data good enough to model value against. Where it does not exist, we will say so rather than invent a baseline.
- A realistic view of what your teams can absorb alongside their day jobs.
- Someone who can say no. A portfolio where nothing is rejected is a list.
What changes
- 01You can fund a first use case on a number rather than a conviction.
- 02Rejected ideas are rejected with a recorded reason, so they stop coming back.
- 03Cost estimates include the parts that usually surface after the budget is set.
- 04Sequence follows dependency and value rather than sponsorship.
- 05You know what has to be true before each one starts.
What it costs
€8,500–19,000 per engagement
All prices exclude VAT.
Questions
How is this different from the AI readiness assessment?
This one asks what AI is worth here. The readiness assessment asks whether this organisation can execute it. If you have never assessed readiness, start there — a ranked portfolio you cannot deliver is an expensive way to raise expectations. Where both are run, the readiness findings feed straight into the feasibility scoring here, and the second engagement is shorter for it.
Will you recommend specific tools?
Where a use case clearly needs a category of tooling we will say so. We are not resellers of any AI platform, and the assessment deliberately stops short of a procurement recommendation so it stays usable whichever way you buy.
What if nothing scores well?
It happens, and it usually means the opportunities are real but the preconditions are not met — most often data. That is a readiness finding rather than an AI one, and it is cheaper to hear it here.
Can you do this for one function rather than the whole company?
Yes, and for a first engagement it is often better. A single function with a motivated sponsor produces a more honest result than a survey of everyone.
Leave with your top three risks documented
Thirty minutes with a senior practitioner. No slideware, no sales engineer.