← All services

AI readiness review

Most failed AI projects did not fail at the model. They failed because the data the model needed was not reachable, nobody could tell whether the output was right, and there was no record of what it did.

Those are answerable questions, and they are much cheaper to answer before a build than during one.

The three questions

RetrievableCan the right records be fetched at the moment they are needed? Data that exists but takes a human to locate is not available to an automation.
MeasurableIs there a set of real examples with known-correct answers? Without one, "is it working" is a feeling.
AccountableWhen it gets something wrong, can you find out why? No audit trail means no debugging and no defence.

What you get

The deliverable

Why we would rather tell you no

An automation built on data that is not ready does not fail immediately. It works in the demo, works for a fortnight, and then quietly starts being wrong in ways nobody catches for a month. That outcome is worse for you than not building it, and worse for us than losing the sale.

If the answer is not yet, you get the list of what would change that. Some of it you can probably do yourself.

How it fits

The review is a fixed fee and takes days rather than weeks. If it comes back as a go and you proceed to a build, it feeds directly into the scope — the evaluation set produced here becomes the one used to watch the automation in production.

Thinking about building something with AI? Tell us the task you have in mind. If the honest answer is that your data is not ready, you will get that — along with what would need to change.

Get your free review