An AI audit that ends with your first agent priced.
Every department has an AI idea, and nobody can say which one will pay off. We count the repetitive work, find the task where an agent pays off first and check the data, systems and GDPR rules behind it, usually within two to three weeks. You leave with the scope of that first agent, the gain to expect and a fixed price, free to build it with someone else.
Starting without one
- A pilot chosen because it looked impressive
- Tools tested one after another, none in production
- A budget with no measured gain behind it
- Compliance questions raised after the build
After the assessment
- The first task, chosen on measured volumes
- Its expected gain, with the assumptions in writing
- A fixed price and a date for the first agent
- Data, access and GDPR checked before any code
What is an AI audit, or AI readiness assessment?
An AI audit, or AI readiness assessment, is a short review of how your company actually works, done before any budget goes into AI. It lists the repetitive tasks, counts them, checks the data and systems each one relies on, and ranks them by what an agent would save. It ends with a decision: the first agent to build, its expected gain and its price.
Two mistakes come up again and again on a first AI project. A company picks the idea that impresses in a meeting over the one that saves hours, or it finds out halfway through the build that the data is not where everyone thought. The assessment is there to catch both before you commit a budget.
Your management, your IT team and your DPO will ask the same questions sooner or later: which data goes where, which model, under which rules, and who is alerted when something goes wrong. The assessment answers them in writing, while changing course still costs nothing.
How the assessment runs, usually in two to three weeks.
Week 1
Map the work
Short interviews with the people who do the work, and a look at real emails, files and screens. Each repetitive task is written down with its volume.
Week 2
Check what it depends on
Data, systems, access, security and GDPR constraints, task by task. Anything that would block a project is flagged before it starts.
Weeks 2 to 3
Rank and price
Each task is weighed on what it would save against how hard and how risky it is to hand over. The winner gets a one-page scope, an expected gain with its assumptions, and a fixed price.
Last day
Decide together
A working session with your management to go through the plan. Every document is yours, whoever builds next.
What you will know for sure at the end.
Every company asks these six questions before a first AI project. Here, each answer is worked out on your volumes and your systems.
Where to start
BeforeEvery department has an AI idea, and every one of them seems urgent.
AfterOne task comes first, chosen on measured volume and gain, and everyone knows why.
What it is worth
BeforeThe gain from AI is a feeling, or a figure from a vendor’s slide.
AfterThe expected gain is calculated on your volumes, with the assumptions written down.
What it will cost
BeforeQuotes that cannot be compared, and the fear of a bill that drifts.
AfterA fixed price for the first agent, with its scope on a single page.
Whether your data is ready
BeforeNobody knows if the data needed is clean, complete, or even reachable.
AfterEach source is checked, and what needs fixing is listed before the build.
What compliance requires
BeforeGDPR and AI Act questions surface late, and stall the project.
AfterData flows, hosting and human oversight are documented for your DPO from the start.
What to leave alone
BeforeTasks get automated that would have been better simplified, or dropped.
AfterThe plan also says where AI is not worth it, and why.
Assessment, pilot or off-the-shelf tool: how should you start with AI?
Where the risk lands depends on how you start.
| Criterion | Readiness assessment first | Pilot straight away | Off-the-shelf tool |
|---|---|---|---|
| How the first use case is chosen | On measured volume and gain | Often the most visible idea | By what the tool happens to do |
| Data and systems | Checked before any build | Discovered during the pilot | Rarely looked at beyond the tool |
| GDPR and AI Act | Documented upfront | Handled late | Depends on the vendor |
| What you have at the end | A scope, a gain and a fixed price | A prototype, sometimes | Licenses |
| Main risk | Low: blockers are known in advance | A pilot that never reaches production | A tool nobody adopts |
| Best for | A first project, or a restart after a stalled pilot | A single, well-known task | A standard need, covered by a mature product |
What decides the price of an assessment.
You get a fixed price for the assessment itself, in writing, before we start. It depends on how much ground there is to cover, listed below. The plan you receive at the end then carries the fixed price of your first agent.
What a first agent costs: market ranges in our guide- 01How many teams and processes we look at.
- 02How many systems and data sources we check.
- 03The compliance requirements of your sector.
- 04How many sites and languages are involved.
- 05On site, remote, or a mix of both.
AI readiness assessment: the questions buyers ask us
How long does an AI readiness assessment take?
Usually two to three weeks, depending on the number of teams and systems involved. The dates are set before we start, and the assessment ends with a working session in which you receive the plan.
What exactly do we get at the end?
A short written plan: the tasks mapped with their volumes, the opportunities ranked, the data and compliance checks, and, for the first agent, a one-page scope, the expected gain with its assumptions and a fixed price to build it.
Do we have to build the agent with you afterwards?
No. The plan is yours, and it is written so that another provider or your own developers could use it. If you do build with us, the work starts straight from the plan, with nothing to redo.
Who needs to be involved on our side?
A sponsor in management, the people who do the tasks we look at, and someone from IT for access to systems. Interviews are short and planned around your teams’ schedule. Your DPO is welcome at the compliance review.
What if the assessment finds that AI is not worth it?
Then you get that in writing, with the reasons. Sometimes a process needs simplifying first, or a tool only needs reconfiguring. You avoid paying for a project that would never have paid off.
Can we start with the online readiness check?
Yes. The online check gives you a first picture in ten questions, in about two minutes. The assessment goes much further: it looks at your actual work and data, and ends with a priced first agent.
A plan you can act on, whoever builds it.
Security and dataYour data stays yours.
We look at samples with you, under a confidentiality agreement if you wish, and you decide what happens to them at the end.
Every gain comes with its assumptions.
Volumes come from your systems or your teams, and each estimate shows how it was calculated.
Honest about what AI cannot do.
The plan names the tasks where AI is not the answer, and says what would work better.
Fixed prices, in writing.
One for the assessment, agreed before it starts. One for the first agent, in the plan.
You own the deliverables.
The task map, the checks and the scope are yours, to use with any team.
Where the assessment usually leads
- AI readiness check Ten questions, about two minutes, before you talk to anyone.
- AI governance Who decides, who checks, and what the AI Act asks.
- How to choose an AI agency The questions to ask any provider.
- AI agents To build the first agent the plan scopes.
- AI integration When the gain is in connecting your tools.
- AI training for employees To prepare the teams who will work with it.
Where would AI pay off first for you?
Tell us which tasks weigh on your teams. We reply within one business day, and a free 30-minute call is enough to see how the assessment would run in your company.