Before you pay for the next AI project, do you know what would make you stop it?
Bring me one decision you actually have to make. A pilot to extend or to close, a contract to sign or not, a plan the group is waiting for.
We spend three hours on it with your executive team, and five working days later you have a one-page memo. In practice the memo ends up covering three decisions, because the one you bring always drags two others behind it, usually who owns the result and what happens with the data or with the vendor.
What you get
One page, with the same ten headings every time: the decision, the verdict, the reasons, the strongest argument against, what is still missing, the condition for stopping, the condition for scaling, who owns it, when we look at it again and what happens this week. I write it so that your board can read it without me in the room. If it has to go to the group, it goes in English.
A sample memo
Sample. The company is invented, AxClaim Group, the one from Călătoria AI, and so are the numbers.
| Decision under review | Extend the claims-triage assistant from the pilot office to all seven regional offices, on the three-year licence the vendor has proposed. |
| Verdict | Run a limited pilot on two more offices first. Do not sign the three-year licence yet, and do not stop the project. |
| Why | The 63% correct-routing rate comes from one office, and the mix of claims there is not the mix in the other six. The 80% the vendor quotes was measured on their own benchmark set, not on your claims. And at the moment nobody in the company owns the routing figure, so nobody would notice if it dropped. |
| The case against this verdict | If the 63% holds on two offices with a different claim mix for sixty days, then scaling is the right call, and every month of delay costs roughly what the six waiting offices spend on manual triage. |
| Missing evidence | The claim mix per office. The cost per handled claim at full volume, including compute, human review and rework, which is a different number from the licence fee. A named owner for the routing figure. |
| Stop condition | Correct routing under 55% on the second office after thirty days, or a running cost above 1.4 times what the vendor projected. |
| Scale condition | Two offices above 60% for sixty consecutive days, a cost per claim below the manual baseline, and an owner in place. |
| Owner | Operations Director. |
| Next review | Day 30 and day 60, in the regular executive meeting rather than in a separate workshop. |
| This week | Put the three-year licence negotiation on hold. Sign a sixty-day extension for two offices at pilot pricing. Name the owner by Friday. |
How it works
- We start with a half-hour call, at no charge. Sometimes the honest answer at the end of it is that you do not need this now, and if so I will say it first.
- You send me what you have: the proposal, the pilot numbers, the contract draft, the board paper. I read all of it before we meet, so that the session starts from your facts and not from my slides.
- Three hours with the executive team, six people at most. You run the meeting. My job is to show where you are not actually in agreement and what that disagreement is costing you.
- The memo arrives within five working days.
- We talk for half an hour once you have read it. After that I do nothing further unless you ask me to.
Price
The fee is fixed: 2,500 euros plus VAT, for an executive team of up to six people. Until 30 November 2026 the first three clients pay 1,950 euros plus VAT.
I am open about the reason for the lower fee. The format is new and I have not run it for a paying client yet. The experience behind it is twenty-five years old; the packaging is a few months old. In return I ask the first three clients for an hour of frank feedback on the method and for permission to quote what we found, without any names.
What this is not for
I do not implement, I do not run vendor selections, I do not give legal opinions and I do not write software. If you need any of those, I can tell you who does them well.
If there is nothing on the table right now, do not book a session. Tell me when there will be, a budget round, a board meeting, a contract coming up for renewal, and I will get back to you then.
Send the decision
Write me three lines: what has to be decided, by when, and who carries it if it goes wrong. If you want to be thorough, answer the questions below. I answer within one working day, usually with two dates for a call.
- What has to be decided? (one sentence)
- By when?
- Who carries it if it goes wrong?
- What is at stake: money, credibility, a contract?
- What would make you stop? (optional, and the most useful question on this list)
- Nothing on the table right now? Tell me when there will be (budget round, board meeting, contract renewal) and I will get back to you then.
Or write directly to stefan@myairadar.com.