Find out whether there is anything here worth automating.
Two weeks watching how the work actually happens, then a shortlist sized in hours and dollars, an architecture for what we would build, and a fixed price for building it. $2,500, credited in full against a build.
The expensive mistake is not picking the wrong tool. It is automating the wrong process.
Most AI projects that fail were correctly built. They automated something that did not cost much, or something whose cost was in the judgment rather than the keystrokes, or something the business stopped doing four months later. The technology worked. The selection did not. Two weeks spent choosing properly is cheaper than two quarters spent building the wrong thing well.
We watch, not askTime-and-motion on the real process, with the people doing it, not a description of it from someone who does not.
Sized in hours and dollarsEvery candidate carries what it costs today and what it would cost to replace.
Including the answer “do not”If nothing here is worth building, that is a finding, and you keep it.Six things arrive. All of them are named.
Artifacts, not activities. Each one is a thing you can open, keep and hand to someone.
A current-state map
How the work actually moves through your business, drawn from watching it, including the steps nobody documented and the workaround everybody uses.
A sized opportunity list
Every candidate process with the hours it consumes now, the share that is genuinely automatable, and what it would cost to build.
A recommendation, ranked
The [two or three] worth doing first and the reasoning, including what we would not touch and why.
A reference architecture
What we would build for the top recommendation: systems, data flow, where a model is doing judgment and where it is not, and where a person stays in the loop.
An honest verdict
An explicit ready, not ready, or not worth it. This is a named deliverable rather than a tone, because an assessment that cannot return a no is a sales call with an invoice attached.
A fixed price for the build
Quoted in writing, valid for [60] days, credited by the full assessment fee if you proceed within 30.
What happens, and when
The clock starts when we have what we need from you, and the first phase says exactly what that is.
Scheduling
We agree who we need to watch and when. The clock starts when the sessions are booked.
You get · a confirmed scheduleObservation
[4 to 6] sessions of roughly an hour with the people doing the work, plus a review of the systems involved.
You get · the current-state mapSizing and architecture
Candidates sized, ranked, architecture drafted for the top one, build priced.
You get · the recommendation and the priceSession
A [60]-minute working session on the findings, with the documents delivered before it, not during.
You get · everything, in your handsWhat it is, and what it is not
What actually arrives
Working documents, not a slide deck. [A 15 to 20 page assessment], a one-page executive summary someone can take to a board, the sized opportunity list as a spreadsheet, and the reference architecture as a diagram.
Pricing
Published. Nothing here is quoted after the fact.
Questions
If your question is not here, one of us will answer it directly.
Still have questions? →A free discovery call is a sales document: it exists to open an engagement, so it cannot conclude that you should not have one. This can, and the verdict is a named deliverable. You keep everything it produces either way.
NextSomething eating
a person's week?
Tell us what it is in two sentences. We will say whether it sounds automatable and whether the assessment is worth it, and if the answer is that a person should keep doing it, that is free advice you can keep.