When the job is unclear, make the decision first.

The AI Reality Audit is a short, fixed diagnostic—not a maturity model. It is for operators with tool sprawl, pilot graveyards, or a queue that has not yet earned a build.

One credible next move—or a documented no.

We trace what is actually in use, where work waits, and whether a product already solves the job. A Sprint is recommended only when the workflow, owner, system, and metric can hold up.

  • InventoryTools, models, vendors, and shadow work already in the operation.
  • Reality checkWhat is live, what is a demonstration, and what is blocking the queue.
  • BoundaryData, security, vendor, product, and regulatory constraints before a build is proposed.
  • DecisionStop, keep, buy, or one Sprint recommendation with a metric and evidence plan.

Start with a Sprint when you can.

If you can already name the recurring queue, source system, accountable owner, and number, the Outcome Sprint is the direct path. A diagnostic is not a gate.

Use the audit when you cannot.

The useful answer may be to configure an existing product, narrow the operating problem, or stop. We will not turn uncertainty into a broad AI programme.

Name what is unclear.

Tell us the queue, source system, and number. We will tell you whether it is a Sprint, an audit, a product you should buy, or a no.

Check the fit