AI readiness audit
One week inside the way your company actually works, and a ranked answer to the question that matters at the start: which process should AI take first?
- Duration
- One week
- Your time
- 4-6 hours total
- Output
- Written 90-day plan
- Lock-in
- None
What an AI readiness audit is
An AI readiness audit is a one-week structured review of how your company works today, ending in a ranked list of the processes where AI would pay back fastest. It looks at the processes, the data behind them, the systems they live in and the people who own them. The deliverable is a decision document with numbers attached, not a strategy essay about the future of work.
Most failed AI projects did not fail technically. They failed because the company started with the tool instead of the process, picked whatever looked most impressive in a demo, and discovered in month four that the chosen process happened four times a month. The audit exists to make that mistake cost one week instead of two quarters.
What you receive
- Process and data inventory. Every process we reviewed, how it runs today, how many hours a week it consumes, and what state the underlying data is in.
- Ranked use case shortlist. Each candidate scored on value, feasibility, data readiness and risk, with estimated hours saved and build effort next to it.
- Build, buy or leave alone. A recommendation per item. Plenty of things are better solved by an existing product, a database view or simply better rules, and we say which.
- Deployment recommendation. Cloud, on-premises or hybrid, argued from your data sensitivity and volume rather than from preference.
- A 90-day plan. Which single process to pilot first, what success looks like in measurable terms, and what has to be true before the pilot starts.
The week, day by day
- Day 1 - kickoff. One call with whoever can approve change. We agree the scope, the departments in play and what a good outcome would look like for you.
- Days 2-3 - interviews. Two or three conversations with the people who actually run the work. Not the org chart version of the process, the real one, including the workarounds.
- Day 3 - systems and data. A look at the ERP, CRM, mailbox and file storage involved. What has an API, what does not, where the data is clean and where it is entered by hand.
- Day 4 - scoring. Every candidate use case goes through the same scoring model, so the ranking is comparable rather than a matter of taste.
- Day 5 - delivery. A written document and a walkthrough call. You leave the call knowing which process is first and what has to happen before it starts.
How we rank a use case
Every candidate gets scored on the same four axes. It is deliberately unglamorous, because the point is to compare an invoice queue against a customer support inbox honestly rather than to pick whichever one demos best.
| Axis | What we measure | Kills the case when |
|---|---|---|
| Value | Hours per week consumed, cost of the errors it produces today | It runs a few times a month |
| Feasibility | Whether the rules can be written down and the systems opened up | Nobody agrees what the rules are |
| Data readiness | Format, consistency and volume of the input the AI would read | The source is paper or free-text chaos |
| Risk | Consequence of a wrong output, regulatory and contractual exposure | A wrong answer is unrecoverable |
What we look for that companies usually miss
- The process nobody mentions. The one somebody has quietly done in a spreadsheet every Friday for six years. It never appears in the brief and it is frequently the best candidate.
- Contract clauses. Your client contracts may forbid passing their data to subcontractors or processing it outside the EU. This constrains the architecture and is far cheaper to find now than after the build.
- Who will own it. A deployed system with no named owner is abandoned within a year. If nobody can be that person, we say so and adjust the recommendation.
- The baseline. If nobody measured how long the process takes today, no result can be proven later. We take the baseline during the audit, because afterwards it is too late.
When you should skip the audit
If you already know exactly which process hurts, the rules are written down and the system has an API, do not buy a week of analysis to confirm it. Go straight to a pilot. The audit earns its money when there are five plausible candidates and no agreement on which comes first, or when a board needs numbers before it will release a budget. Not when the answer is already obvious.
What happens after
The document is yours, written so any competent supplier could execute it. Most clients move straight into a pilot on the top-ranked process, which takes a further two to four weeks and runs with a human approving every output. Some take the document to their own IT team. Both are fine outcomes, and both beat spending a year on an AI strategy that never touches a real process.
Questions we get asked every time.
What is an AI readiness audit?
An AI readiness audit is a short, structured review of how a company actually works today, ending in a ranked list of the processes where AI would pay back fastest. It covers the processes, the data behind them, the systems they run in and the people who own them. The output is a decision document, not a strategy essay.
How long does the audit take and what do we have to do?
One week. From your side it costs roughly four to six hours in total: a kickoff call, two or three interviews with the people who run the processes, and read-only access or screenshots of the systems involved. We do not need your production data to produce the ranking.
What do we actually receive at the end?
A written document with four things: an inventory of the processes we reviewed, a ranked shortlist of AI use cases with estimated hours saved and build effort for each, a build-versus-buy recommendation per item, and a 90-day plan naming which single process to pilot first and why.
Are we obliged to build anything with you afterwards?
No. The document is yours and it is written so another supplier could execute it. We would obviously like to build the pilot, but an audit that only makes sense if you hire us is not an audit, it is a sales meeting with a cover page.
What if the audit finds nothing worth automating?
It happens, and we say so plainly. Some companies are too small for the overhead, some have processes too irregular to model, and some need their data cleaned up before AI is even a sensible conversation. Being told that in week one is considerably cheaper than discovering it in month six.
Keep reading.
Find out what AI is worth to you before you spend anything on it.
Thirty minutes, no deck. We will tell you whether an audit would find anything worth acting on, or whether you should skip straight to a pilot.
Book a free 30-minute call