1. Home
  2. AI and process
  3. AI readiness check

We rolled it out. Hardly anyone uses it.

The model is fine. What it runs on was never written down: who owns the workflow, what a number means, who checks the answer.

AI readiness check7 questions2 minutesNo email

Did your AI rollout stall? Seven questions, one reading.

Think of the one job you most hoped AI would take on. Pick the answer closest to what has happened since. The base on the right fills in as you go.

1 of 7
Where does AI stand in that job today?
Before the rollout, who signed up to a result it should deliver?
When AI gets a case it cannot handle, what happens?
When an AI answer is wrong, who decides what happens next?
When an AI figure and a finance figure differ, which counts?
Could AI answer what people usually ask your two most experienced colleagues?
What happens to what AI produces?

Your reading

 

Talk it through, 15 minutes →
What you saw

Why it likely stalled

A first step you can take

If more is needed, the optional Decision Diagnosis maps it, at a fixed fee.

Seven questions. No email. Your answers stay in your browser.

It usually looks like this.

From real work. Anonymised, details changed. Some merge two situations.

Owners and definitions before automation.

AI built the tools. Nobody ran them.

UnderneathEach tool kept its own copy of the facts and ran on one person's account.

What changedOne list of which system owns which fact, and each open question written as a decision with an owner.

How it works.

The same four steps, whether it is one workflow or the whole company.

Start: an intro call.15 minutes, free, person to person.
  1. Decision Diagnosis

    Find the real problem.

    Not the symptom on your desk, the cause underneath it.

    On its own, this step is the Decision Diagnosis: optional, a fixed fee, and you can stop there.

  2. Then, if you want it

    Design it with you.

    Who decides what, who owns which number, and the rhythm that checks it.

  3. Build and run it.

    With your team, on live data, while the company keeps running.

  4. Hand it over, step out.

    It counts as done only when its owner uses it. Then it runs without me.

On its own, this step is the Decision Diagnosis: optional, a fixed fee, and you can stop there.

Then, if you want it

The same four steps, on the workflow where AI stalled.

Same model. Different base.

Decisions and ownership first. The data follows from that. A capable model on a mapped, owned workflow with defined numbers, written know-how and written limits does useful work.

See it switch on, one layer at a time →

Good workflows to check

  • Month-end close and reconciliations
  • Forecasting and variance tracking
  • Order to cash, and its exceptions
  • Customer questions and escalations
  • Onboarding and access

Bring your answers to a conversation.

Fifteen minutes, free. We talk about your company, and you hear plainly whether going further makes sense.

Luka B. Memon