Governed AI Readiness Assessment

Who this is for

Organizations with more candidate AI use cases than clarity about which ones can actually be built, and a governance function that is being asked to approve things nobody has fully specified.

• Executives want AI value, and IT and security need controls before anything moves.
• Your data cannot leave your governance boundary, which rules out several vendor answers.
• Several departments are proposing use cases and nobody is comparing them on the same basis.
• Nobody can say what any of it would cost to run once it is live.
• A previous pilot stalled and it is not clear whether the obstacle was technical, legal or organizational.

What the assessment covers

1. The use-case portfolio

   Every candidate collected and written to the same standard, so they can be compared. What the pain point is, who owns it, what it costs today, and what a solved version would have to do.

2. Data reality

   Where the data actually lives, who owns it, its quality and lineage, and what your governance boundary permits. This is where most AI plans meet their real constraint, and it is better met early.

3. Risk tiering and controls

   Each use case assigned a tier, with the human-in-the-loop controls that tier requires. Low-risk retrieval over published material behaves very differently from anything touching an individual's record or leaving the organization as a reply.

4. Infrastructure fit

   What your existing estate can carry and what it cannot. Hardware and GPU sizing, operating systems, databases, identity, networking and security, assessed by someone who has been accountable for running estates at scale rather than for selling into them.

5. Cost and total cost of ownership

   What each use case costs to run, modelled per request rather than per seat, including the architectural choices that move the number. Whether models are reached in-house over direct provider interfaces or through an intermediary changes this more than almost anything else.

6. Sequence

   Which use cases go first, and why. Usually the one where the pain is real, the data is available, the risk tier is manageable, and the owning department already wants it solved.

What you get

• A ranked use-case portfolio, each entry specified with acceptance criteria.
• A risk tier and control set per use case.
• A data assessment covering ownership, quality, lineage and boundary constraints.
• A reference architecture, carried through to a procurement-ready bill of quantities where you need one.
• A cost and total cost of ownership model you can change assumptions in.
• A target operating model and governance approach for running this beyond the first use case.

This is delivered as an interactive multi-scenario model rather than a slide deck. Change an assumption and the resourcing, the cost and the sequence move with it. Scope and commercial terms are agreed per engagement.

What this is not

It is not a maturity score, and it does not end with a grade. It is also not a technology selection exercise on behalf of a vendor. If you already know the one workflow you want solved and the question is whether it can be built, skip this and go straight to the AI Pain-Point Prototype Sprint, which answers that question by building the thing.

Why this works

The governance model described here is the one that carried ten AI workflows into production across nine business functions, with 367 governed documents behind a single role-aware gateway, human-in-the-loop controls set per risk tier, and per-query cost metering from the first day.

Read the case study

Start

A useful first message names the use cases already on the table and the constraint you expect to be hardest.