Service 01

AI readiness assessment

Two to three weeks, fixed fee, and a written answer to the only question that matters at this stage: where will AI actually pay off here, and what would it take to do it safely?

What this actually is

Most organisations do not have an AI problem. They have a prioritisation problem wearing an AI costume: fifteen suggestions from around the business, no agreed basis for comparing them, and a nagging sense that the compliance position on most of them is unclear.

This engagement produces the missing basis. We look at what you actually run, what data you actually hold, what constraints you are actually under, and we come back with a ranked list, including the items that should be struck off it.

What we look at

Your existing AI footprint

Everything already in use, approved or otherwise. This routinely turns up material AI processing that nobody has assessed, because it arrived as a feature of a tool bought for another reason. For organisations in EU scope, this is also where prohibited-practice exposure tends to hide.

Your data, honestly classified

For each candidate use case we establish the highest classification of data that will touch the system in production, not in a pilot on sanitised extracts. This single table decides most of the architecture, and it is the step whose omission causes projects to die at the information governance gate months later.

Your systems and data quality

Where the relevant information lives, what condition it is in, whether it can be accessed programmatically, and how permissions work in the source systems. Retrieval quality is mostly a content problem: an organisation whose documents are 40% superseded will get confidently wrong answers regardless of which model it buys.

Your controls and obligations

Which regimes apply, what your DPIA position is, what your regulator will expect to see, and what a procurement panel or assurance board will ask. We would rather establish the evidence requirement in week one than discover it in month nine.

Your people

Who would own this, whether they have capacity, and whether the process you want to automate is actually understood well enough to specify. The most common technical blocker we find is not technical: no one internally owns the outcome.

How it runs

Week 1: Discovery. Kickoff, document and system review, four to six interviews with the people doing the work rather than the people describing it. Inventory of current AI use.

Week 2: Analysis. Data classification, opportunity sizing, constraint mapping, and a technical feasibility check against a real sample of your material rather than a hypothetical one.

Week 3: Report and session. Written report, then a working session with your team to walk through the recommendations, argue with them, and agree what happens next.

Who it is for

Organisations with several competing AI ideas and no agreed way to choose between them.
Teams that need a costed business case before anyone will release budget.
Regulated organisations that suspect their compliance position is the real constraint and want it established before, not after, they commission a build.
Boards that have been shown an impressive demo and want an independent read on whether it survives contact with production.

What it is not

It is not a strategy deck. We are not going to spend three weeks telling you that AI is transformative and that you should establish a centre of excellence. The output is specific: these use cases, in this order, on this architecture, at roughly this cost, with these risks owned by these people, and these ones you should drop.

Questions we get asked

What makes this different from a free AI readiness assessment from a vendor?

A vendor assessment is a qualification exercise. It is free because its purpose is to establish which of the vendor’s products you should buy, and it will not conclude that you should buy none of them. We charge for this work precisely so that "do nothing here" stays available as a recommendation, and it is one we make regularly.

We already know what we want to build. Do we need this?

Often not, and we will tell you. If you have a specific, well-defined use case, you know what data it touches, and you have a view on the deployment constraint, skip straight to implementation and we will scope that instead. This assessment earns its fee when there are ten competing ideas and no agreed basis for choosing between them.

How much of our time does it take?

Roughly six to ten hours across the engagement, mostly in weeks one and two: a kickoff session, four to six interviews with people who actually do the work, access to a sample of representative documents or data, and a review session on the draft. We do the rest.

Will you look at the tools we are already using?

Yes, and this is consistently where the surprises are. Most organisations underestimate their own AI footprint substantially, because it arrives embedded in tools procured as something else: CRM lead scoring, HR CV screening, service desk triage. Some of that carries more risk than anything on the wish list.

What if the answer is that AI will not help us?

Then that is the report, with the reasoning set out so you can use it with your board. It happens. A meaningful proportion of proposed AI projects are better solved by a database query, a fixed rule, or fixing the process that generates the work, and finding that out for the cost of an assessment is a good outcome, not a wasted one.

Can this feed a business case or funding bid?

Yes, it is designed to. The output includes costed options, quantified benefit estimates with the assumptions stated, and a risk register, which is the substance of most internal business cases and public sector funding submissions.

Start with a straight answer

A 30-minute call, no pitch deck. Tell us what you are trying to do and we will tell you whether AI is the right tool, what it would take, and what it would cost, or that you should not bother.

Book an intro call See how we price work