In one paragraph
One workflow, one tool, one number. We find the single place in your business where AI would genuinely carry a step better than the alternative, set it up properly with a human accountable, and measure that workflow's own number before and after. If there isn't a place worth doing, we tell you, and that is a real possible outcome.
AI where it pays, nowhere else
"Everyone says we should be using AI. Nobody can tell me for what."
What is actually happening
You are being sold a category, not a fix.
The pitch is almost always framed as adopting AI, which is not a business outcome any more than adopting electricity was. Underneath it there may well be one workflow in your business where a tool would genuinely carry a step better than what you do now. Finding which one, and proving it afterwards, is the entire job — and it is the part nobody sells, because the answer is sometimes "none of them this year."
What we see in practice is that the AI already in these businesses is individual and cosmetic: someone drafting, someone making a flyer, someone curious. Useful, occasionally. Not connected to how the business runs, and not something anyone has measured.
One workflow, measured. Not a category.
- Adopting a category
- Fixing one workflow with a number on it
How this gets sold elsewhere
Consultancies sell a readiness assessment, which measures you against a journey you did not ask to be on and ends in a roadmap.
Vendors sell seats, and the seat is the outcome from their side.
Agencies sell pilots, which are structured so that the pilot succeeding is the deliverable rather than the business improving.
None of those is dishonest. All of them are paid whether or not the thing pays, which is why the second measurement is so rarely taken.
What we do
We start from the work, not the tool. Which step is slow, repetitive, dependent on one person, or produces something a customer waits for.
We shortlist the candidates and pick one. One, because the whole argument of this page is that a measured single change beats an unmeasured programme.
We take the baseline before anything changes. If we cannot name a number for it, that is a signal the workflow is the wrong candidate.
We set it up with what you already pay for, wherever possible. A great deal of what gets sold as new capability is already inside a subscription you have.
We put a human at the gate. Anything that reaches a customer, a patient or a filing is checked by a named person before it goes, and that check leaves a record.
We measure the same number again.
And if the honest answer is that it did not pay, we say so, and we say what to do about it.
What you get
The shortlist, with why each workflow was or was not a candidate.
One workflow, set up and running, with a named person accountable at the gate.
The rules for it — what may go in, what may never go in — consistent with the ground rules, where those are in place.
The before-and-after number, on the same definition.
A plain statement of whether it paid.
What we measure
That workflow's own number. Whatever the work itself is judged by — time to produce, error rate, share needing rework, time a customer waits. One number, taken twice, on a definition agreed before we start.
An illustrative example — not a client
Illustration.
Four candidate workflows. Two were rejected because nobody could name a number for them, one because the information involved could not go into any tool the business could properly control, and the fourth was taken forward.
The measured result was a reduction in the time a customer waited for a written answer. The tool was already inside a subscription the business had been paying for since the previous year.
The most useful sentence in that engagement was the one about the rejected candidate.
What this is not
Not an AI strategy, and not a readiness assessment. One workflow.
Not automation that removes the person. The gate is a person, on purpose, and on anything that reaches a customer it stays a person.
Not permanent. It ends with a number and a written setup you own.
And not something we will sell you if the foundations are not there. If your phones do not reach a person and two of your systems do not talk, AI is not your problem this year, and we will say so.
If you are a law firm, a medical practice or anything regulated
The rules that apply to you are about the work, not the tool. A human checks anything that carries your signature or reaches a patient, and the check leaves a record. We write that into the workflow rather than into a policy, because the policy is not what fails. Your attorney, carrier or privacy officer verifies and signs.
What this costs
We do not publish a price, and here is the reason rather than an excuse: we would be guessing. Two businesses with the same symptom are frequently a factor of five apart in what the work involves, and a published number would either scare off the smaller one or shortchange the larger one.
What we will commit to:
The quote is fixed before any work starts, and it is a single number, not an hourly rate that grows.
We name it after we have seen the work, not before. Which means the first conversation costs you nothing and commits you to nothing.
If we think something is not worth doing, we say so — including when what you are asking for is the thing we would be paid to do.
Nothing renews automatically. Every engagement has an end.
Every number on this page is labeled — independent research, vendor data, or your own measurement. → How we use numbers
→ And afterwards: Measure It Again.
Start with a conversation. No proposal until we have seen the work.
Start a conversation