The Operator's Diagnostic

Three days inside your team. Then a straight verdict, and it can be "do not build yet."

I watch how the work actually happens in your finance and operations function, not how it is supposed to happen. You get a ranked shortlist of what an agent could take over, what it would take, and what has to be fixed first.

Book 30 minutesWhat you get
$18,500
fixed, all-in
3 days
on site, across two calendar weeks
~3 weeks
end to end, readout included
9
artifacts, and every one of them is yours
01How it runs

Watching the work, not interviewing about the work.

On site, 3 days

Shadow the work

  • Four to six shadow sessions
  • A session with IT on the question that decides everything: can an agent read from your systems and write back reliably
Within a week

Written findings

  • Process maps, the shortlist, the value model
  • A verdict on whether your data can carry any of it
Then

Executive readout

  • 90 minutes with you and whoever owns the number
  • The recommendation, including "do not build yet"
02What you get

Nine artifacts. Yours to take anywhere, whether you continue with me or not.

Maturity readWhere your function actually sits, and whether agents are the right next move at all
Process mapsEight fields per workflow: trigger, steps, systems, handoffs, knowledge sources, decisions, delays, exceptions
Workflow shortlistThree candidates, ranked, each with a baseline number in your units and a named owner
Systems readiness verdictWhether your data can support agents, or whether month one is plumbing
Boundary cardFor the top candidate: what it may read, change, escalate, and never do
Acceptance testsSix tests minimum, written before anything gets built, graded pass or fail
Value modelWhat the work costs today, what is recoverable, and what cannot be sized
Executive readoutA 90-minute session with you and whoever owns the number
The recommendationIncluding permission to say do not build yet
03What I need from you

One named owner, and access to watch.

From your side

  • One named owner for the diagnostic itself. If nobody is assigned, that is the first finding.
  • Room for four to six shadow sessions: I sit beside the people doing the work.
  • Someone from IT for one session.

What this is not

  • Not a readiness score. A score tells you where you rank. This tells you what to do.
  • Not a tool recommendation. I take no referral fees from any vendor.
  • Not a pilot. Building comes after, and only if the findings support it.
04Three questions I ask everyone

Asked the same way, of everyone. Which one people reach for tells me more than the answer.

  1. How long does it take you to get a straight answer on a number you already report every month?
  2. What only runs right now because one specific person has not left?
  3. How many AI tools are people using that you did not approve, and is any of it making the work faster?

Want to ask yourself first? Take the five-minute checklist: is your AI already paid for?

05After the diagnostic

Each step is papered separately. Expansion has an entry condition, and I will tell you when it is not met.

If the verdict is build

Build, with your owner: one or two days a week, with the person who will run it after I go. You own what gets built.

Once it runs

Run, measured: priced per agent in production, not per day of my time, against the baseline the diagnostic set.

If the verdict is not yet

You keep the findings and the fix list. If your existing platform already does the thing, that is what the findings will say.

$18,500
Fixed, all-in. Three days on site, written findings, the readout.

Thirty minutes first.

The first thing we settle is whether your function is even at the stage where this pays. If it is not, I will tell you on that call.

Book 30 minutesSee the record