When the strategy is finished and nothing has changed
Many organisations now have an AI strategy or an opportunity list. It names a dozen promising uses: summarising contracts, drafting customer replies, forecasting demand, answering policy questions. A few months on, daily work looks the same. The list was not wrong. It simply never asked the questions that decide whether an idea can actually be piloted next month: who owns it, can the tool reach the data, what does “better” mean, and who has to approve it.
This engagement asks those questions about your specific ideas, inside your organisation, and gives a sponsor something concrete to fund or decline.
Commissioned investigation, not a questionnaire
There are two ways to look at readiness on this site, and they are different purchases.
The free AI readiness assessment is a self-serve tool. You answer questions about your organisation and get an indicative view on the page, without giving an email address. It is a good first step, and many people need nothing more.
This page is a commissioned investigation. I examine named workflows in your organisation: I talk to the people who do the work, look at real inputs and outputs, and check data routes with your IT and data protection teams. The output is evidence about your workflows, not a general score. Neither is an audit or a certification.
What the investigation tests
Each candidate is framed as a specific workflow (“first-draft responses to supplier queries in procurement”, not “AI for procurement”) and tested against the same questions:
- Owner. Is there a named person who will run the workflow in daily work and be accountable for its output?
- Data route. Can an approved tool reach the inputs the work needs, within your data policy, without copying information somewhere it should not be? What approvals are outstanding?
- Baseline. Can we measure how the work performs today, in time, quality or rework, so improvement can be shown?
- Review requirement. Who must check the output, and is that reviewer available and willing?
- Value and volume. Does the work recur often enough, and matter enough, to justify the change?
Ideas that fail a test are not discarded. The report records what would make them ready, so they can come back later.
The signature deliverable, illustrated
You receive an evidence-backed shortlist, access dependency map and funded pilot brief. Illustrative extract from a shortlist:
| Candidate workflow | Owner | Data route | Baseline | Reviewer | Verdict |
|---|---|---|---|---|---|
| Supplier query first drafts (procurement) | Procurement lead, confirmed | Shared mailbox export approved for pilot | 3 weeks of logged queries | Senior buyer | Ready: pilot brief written |
| Policy Q&A for staff (HR) | HR operations manager | Policy library in document store; permissions need review | Ticket volumes available | HR policy owner | Ready after permission review |
| Demand forecast commentary (operations) | Not identified | Warehouse access request pending | None | Unclear | Not ready: needs owner and data access |
Illustrative example. It shows the format, not a client’s assessment.
The access dependency map shows, for each shortlisted workflow, the systems, approvals and people involved and which are outstanding. The pilot brief for the top candidate sets out scope, owner, user team, baseline method, measures, review points, data rules, risks and stop criteria, written so a sponsor can approve it and any competent team could run it.
How acceptance is judged
The sponsor accepts the investigation when every candidate has been assessed against the same tests with recorded evidence, the dependency map has been checked by IT and data protection, and the pilot brief is complete enough to fund or decline without further work. A conclusion that nothing is ready yet, with the reasons, is an acceptable result.
Ownership and handover
The sponsor owns the shortlist and the funding decision. The named workflow owner owns the pilot brief. If you fund the pilot, you can run it internally, use the free pilot scorecard to judge it, or commission it as an AI enablement pilot. If the main blocker is data or integration rather than ownership, see make data and integrations ready for AI.