AI Readiness Score

Why this question matters

Interest in AI is not the same as readiness to use it well. A business needs a defined purpose, suitable information, responsible people and operating controls before an AI tool can support reliable work. This scorecard examines twelve statements across those four areas. It helps identify the conditions needed for a bounded use case. It does not rank technical sophistication or recommend replacing people with a system.

Begin with one concrete use case. Drafting an internal summary, helping classify enquiries and recommending a consequential decision have different requirements. State what the output will be used for, who will review it and what happens if it is wrong. Measure the current process where possible. Without a baseline for time, cost and quality, it becomes difficult to tell whether the new tool improves the work or simply makes it look different.

Review the information required by the use case. Check accuracy, permission to use it and the handling of sensitive material. A technically accessible document is not automatically appropriate to send to an external service. Consider what the provider receives, who can access the result and how the work continues if the service is unavailable. Keep credentials and confidential records out of an exploratory test until the required arrangements are understood and authorised.

Understand the method

1. Twelve ratings across four dimensions are normalised to 0 to 100 and averaged.

The twelve ratings are converted from one to five into values from zero to one hundred and averaged. Each of the four dimensions contains three statements: purpose, information, people and controls. The score expresses your assessment of those foundations. It does not inspect systems, verify permissions or test the quality of an AI output. A higher score indicates more established preparation under the stated rubric.

Read the weakest dimension as a missing operating condition. Weak purpose means the business may be adopting a tool without a clear problem or baseline. Weak information means quality or permission may be unresolved. Weak people arrangements mean review responsibility, skills or fallback work is unclear. Weak controls mean evaluation, spending limits or incident handling needs attention. Some gaps should stop a particular pilot even when the overall average looks strong.

Build a small evaluation set of representative tasks with acceptable outcomes defined in advance. Include difficult cases and situations where the correct response is to ask a person rather than invent an answer. Assign a reviewer who understands the work, record errors and keep a practical manual route. Set limits on use and spending. Expand only when the evidence supports the specific use case, and repeat evaluation when the model, instructions or source information changes.

Keep the result in perspective

This is a transparent planning calculation or self assessment, not a sector benchmark, professional valuation or a prediction of an outcome.

The scorecard is an educational readiness rubric, not a security audit, legal assessment or AI certification. It applies no external benchmark. Self reported ratings cannot establish that personal information is handled lawfully or that the system is safe for a consequential decision. Use relevant professional and technical review where the proposed activity requires it.

AI performance can vary with inputs, instructions, model changes and the task itself. Successful examples do not prove that all future outputs will be correct. Human review must be meaningful: a person who lacks time, context or authority to reject an output cannot provide a reliable control simply by being named. The tool also does not calculate financial savings or determine the appropriate staffing structure.

Read the full limitations or explore how Ayodeji approaches this work.