An AI readiness assessment asks a more useful question than “Are we using AI yet?” It asks whether the business has a specific problem, a workable process, usable information, accountable people and a way to measure whether AI improves anything.
How the assessment works
Choose one potential AI workflow before scoring. Do not score the entire company in the abstract. A business may be ready to automate meeting follow-up but nowhere near ready to automate a sensitive customer decision.
AWS's SMB readiness guidance similarly starts with a defined business outcome, data sources, ownership and measurement rather than technology alone. See AWS's AI readiness checklist.
1–3: Business problem
- We can name the specific business problem. Not “we need AI,” but something like slow proposal turnaround or repetitive support triage.
- The problem happens often enough to matter. There is meaningful volume, cost, delay or inconsistency.
- We know what better looks like. Faster response, fewer manual steps, higher consistency or another observable outcome.
4–5: Workflow
- The current workflow is documented well enough to explain. Inputs, steps, decisions and outputs are understood.
- We know where human judgment must remain. The pilot is not an excuse to automate consequential decisions blindly.
If these are difficult to answer, first use the AI opportunity assessment framework.
6–8: Data and systems
- The information the workflow needs is available. Documents, records or knowledge are accessible and reasonably current.
- We know who owns that information and who may access it.
- The systems involved can support a practical pilot. Through exports, approved integrations, APIs or a contained manual test.
Perfect data is not required. But if nobody knows which source is authoritative, AI may simply make confusion faster.
9–10: People and ownership
- A business owner is accountable for the outcome. Someone can decide whether the result is actually useful.
- The people doing the work are involved. The pilot is not being designed entirely around them without their input.
11–13: Risk and governance
- We know what sensitive information may enter the workflow.
- We have decided where human review is required.
- We can limit access and test safely before expanding.
This matters more as systems move from drafting content to taking actions. Permissions, logging and escalation should grow with autonomy.
14–15: Measurement
- We have a baseline. We know roughly how long, how much or how well the process performs today.
- We have a pilot success measure and a stop condition. We know what would justify continuing—and what would tell us to stop.
What your AI readiness score means
| Score | Interpretation | Next move |
|---|---|---|
| 0–10 | Foundation first | Clarify the workflow, information, ownership and outcome before buying technology. |
| 11–20 | Promising but incomplete | Close the weakest readiness gaps and design one contained pilot. |
| 21–30 | Potentially pilot-ready | Define scope, guardrails, baseline and a short test. Readiness does not remove the need for validation. |
What to do next
If your score is high, resist the urge to launch five projects. Choose one of the AI use cases for small business, write down the baseline and run a bounded pilot. If the opportunity is strategically important or crosses several systems, an AI strategy engagement may help prioritize the roadmap.
If the score is low, that is useful information too. Improving the process, source data or ownership may create value even before AI enters the picture.