Find out whether AI is your answer before you fund it
Most AI work fails on the problem, not the model. Two weeks of looking is cheaper than a year of building the wrong thing.
Anyone can build the demo. We build what runs after. The diagnostic exists because the expensive mistake is not a bad model. It is six months spent on a problem that never needed one.
The failure is usually upstream of the model
RAND studied why AI projects die. The leading cause is not technical. It is teams solving a problem the business did not have, or one the data could not support.
So the diagnostic starts with your workflow and your data, not with a model choice. By the end you have a scoped plan, or a clear reason not to build.
- 01What the system must get right, and what it may get wrong.
- 02Whether your data can support it, honestly assessed.
- 03The failure that would cost you most, and the checkpoint that catches it.
- 04What it takes to keep running after launch.
What we bring
22
production systems delivered
2
weeks, fixed scope
1
outcome: a plan, or an honest no
0
dollar figures on this site
How the diagnostic runs
- 01
Walk the workflow
We follow the work as it happens today, with the people who do it. Most of what matters is visible here.
- 02
Test the data
We check whether the data can support the decision you want automated. This is where most ideas fail, and it is better to fail here.
- 03
Name the failure
What happens when the system is wrong? That answer decides whether a human checkpoint is required and where it goes.
- 04
Scope or decline
You get a plan with a fixed scope, or a written explanation of why we would not build it.
- ObserveThe workflow as it runs.
- AssessCan the data carry it?
- ModelWhat breaks, and how badly.
- ScopeA plan, or a no.
The honest no is the outcome nobody sells, and the one that saves the most money.
Systems that started this way
“They will treat your vision like their own and build it that way.”
Ron Klabunde · Founder, SmartREI ↗
A workshop against a diagnostic
| Criterion | A discovery workshop | A paid diagnostic |
|---|---|---|
| Output | A slide deck and enthusiasm. | A scoped plan, or a written no. |
| Data | Discussed. | Tested against the actual decision. |
| Failure | Rarely raised. | Named, with the checkpoint that catches it. |
| Incentive | Ends in a proposal. | Can end in advice not to build. |
Common questions
01What if the answer is that we should not build it?
Then we write that down and explain why. It has happened, and it is the most valuable outcome we sell. A vendor whose diagnostic always concludes in a build is running a sales process.
02Who from your side actually does this?
Senior engineers who have shipped production AI, not a strategy team who will hand off. The person assessing your data is the person who would build against it.
03How is this different from an AI readiness assessment?
A readiness assessment scores your organisation. A diagnostic scopes one problem. If you want the score, our AI readiness tool is free and takes four minutes.
04Do we own what comes out of it?
Yes. The plan, the data assessment and any prototype code are yours whether or not you build with us.
