AI readiness assessment

AI work starts with finding the data. Unknown or scattered sources are the most common early delay. Weight 2.

Data locked in screens or PDFs needs extraction work before any model can use it. Weight 2.

Models repeat the gaps and errors in their inputs. Weight 1.

Examples are needed to test whether the AI is right, and later to tune it. Weight 1.

Customer contracts and privacy law can restrict sending data to a model provider. Weight 2.

You cannot automate a process nobody can describe. Writing it down often finds the real bottleneck. Weight 2.

Automation pays back through repetition; rare tasks rarely justify the build. Weight 1.

Without a baseline you cannot show the AI helped. Weight 2.

AI output is probabilistic; a review or hand-off step keeps mistakes contained. Weight 2.

Projects without an owner stall between teams. Weight 2.

Integrations, evaluations and monitoring need ongoing engineering after launch. Weight 1.

They know the edge cases, and adoption fails when they are left out. Weight 1.

Model API costs scale with use. Price them early with the cost calculators. Weight 1.

The cost of an error decides how much review and testing you need. Weight 2.

Some uses are restricted or need documentation before launch. Weight 2.

Prompts and outputs can leak data across users if access is not designed in. Weight 1.

Evaluations catch regressions when prompts, data or models change. Weight 1.