Decisions & strategy
An AI readiness checklist you can actually use.
You do not need an “AI maturity score” to start a useful conversation. You need enough evidence to decide whether one operational problem is ready for a scoped next step.
The practical answer
Prepare one workflow, observed examples, a baseline, authorised data, an accountable owner, unacceptable failure cases and a realistic operating budget. Missing items are questions to resolve, not a reason to buy a larger project by default.
- Describe
A recurring task and its current friction.
- Evidence
Examples, access, baseline and exceptions.
- Decide
Proceed, investigate a gap or stop.
1. Describe one workflow and the current baseline
Choose a task that has a recognisable start and finish. “Use AI in operations” is too broad. “Turn an approved client intake form into a complete project record” is specific enough to inspect. Walk through a recent ordinary case and a difficult one with the people who do the work.
- Trigger: what starts the task, and in which tool?
- Output: what must exist when the task is complete?
- Baseline: what volume, handling time, waiting time and rework have you observed?
- Alternative: could a form, a rule or an existing feature solve the same problem?
2. Check the data and access before choosing a model
Gather a small, representative set of examples that you are authorised to share. Include missing fields, duplicates, poor formatting and cases that should be rejected. Remove unnecessary personal or confidential information. Synthetic examples are useful for explaining the flow, but do not prove real-world accuracy.
List each source and destination tool, who owns access, and whether a test environment is available. Confirm data handling, retention and provider settings with the responsible people before using real records. A model choice cannot fix missing permission or inconsistent source data.
3. Name the owner and define acceptable output
Assign one person to the business outcome and one to the technical operation; they may be the same person in a small team. Identify who can approve data access, review ambiguous output and stop the workflow if it behaves unexpectedly.
- Which fields must be correct, and which must never be guessed?
- Which decisions require human approval before an external action?
- Which failures are tolerable, and which must block processing?
- How will you test duplicate events, incomplete inputs and a failed integration?
- What evidence will the business owner review before accepting delivery?
4. Plan the operating budget and handover
Budget the people and software needed after launch, not just the build. Set an owner for usage monitoring, a route for reporting problems and a manual fallback. Document who controls accounts, configuration and access when a supplier relationship ends.
Write the first release small enough to evaluate: one process, one accountable team and a defined set of test cases. More channels and integrations can come after the first scope meets its acceptance criteria.
5. Copy this brief into your project notes
Complete the prompts below in plain language. Mark unknowns explicitly. This is a conversation aid, not a validated maturity test or a guarantee that AI is the right answer.
| Prompt | Your project note |
|---|---|
| Problem and boundary | We want to improve … The workflow starts when … and ends when … |
| Evidence | Our current process takes … We observed … Exceptions include … |
| Access and ownership | The data owner is … The process owner is … The reviewer is … |
| Acceptance and limits | A successful case must … The system must never … If it fails, we … |
| Budget and decision | Delivery and operating limits are … The next decision is … |
Bring examples that can disprove the proposal
Prepare one ordinary case, one case with missing information and one conflicting case. Write the expected action for each before testing. These examples start the evaluation; they are not sufficient evidence for a production launch on their own.
AI readiness brief (CSV)
Download the table shown above. Replace illustrative records with authorised examples from your own workflow; this is a working template, not an assessment of your business.
AI readiness brief (CSV)References and further reading
Entrepreneur and AI consultant, I help service businesses improve their operations. I turn software choices into practical plans. My team at Klicky builds what we decide.
Khalid Boudrari