Your business does not need perfect data or a large engineering team to begin evaluating AI automation. It does need a process that someone can explain, access to the information required for that process, and a practical way to judge the result.
An AI readiness assessment should identify a feasible starting point and the work needed before implementation. Its value is in the decisions it enables, rather than a generic maturity score.
Choose one process to assess
“Automate our operations” is too broad for a useful first assessment. Select a repeated task with a recognizable beginning and end: preparing a response, extracting information from a document, routing an enquiry or assembling a draft record.
Ask the person who performs the task to walk through several recent examples. Include one that went smoothly and one that required extra judgment. Differences between those cases often reveal the real scope of the work.
The process needs an owner who can answer business questions and accept the result. Technical access alone cannot resolve disagreement about which outcome the business actually wants.
Check whether the necessary information exists
A team may have many documents while still lacking an authoritative answer to common questions. A price may exist in several systems. An operational policy may be known only to one experienced employee. These are useful assessment findings.
I would identify where each required fact comes from, who maintains it and how often it changes. Then inspect whether the proposed automation can access it in a usable form. A spreadsheet can be a valid source for a limited pilot if its ownership and update process are clear.
The assessment should distinguish absent information from information that is merely difficult to retrieve. Those problems lead to different next steps.
Review five readiness areas
| Area | Evidence to look for | If it is missing |
|---|---|---|
| Process | Representative examples and a named owner. | Map the work before selecting a tool. |
| Data | Approved sources with known update responsibility. | Resolve source gaps or narrow the task. |
| Access | A permitted way to read or update the required system. | Inspect the integration before estimating the build. |
| Baseline | Current effort, delays and correction needs. | Collect a small operational sample. |
| Ownership | Someone available to review results and handle exceptions. | Assign the operating role before launch. |
Measure the current process honestly
A baseline does not have to be an elaborate reporting project. Record how work arrives, how long people spend handling it, where it waits and what causes rework. Separate active handling time from total elapsed time.
If a task takes several days because it waits for a missing customer detail, making the first draft faster may not remove the main delay. The assessment should identify that dependency before promising a large improvement.
Also record which cases should stay with a person. A useful automation opportunity can be the repeatable portion of the process, even when judgment remains necessary elsewhere.
Finish with an action, not a score
I would expect the assessment to recommend one of three practical paths: proceed with a bounded pilot, complete specific preparation work, or choose a different process. Each recommendation should explain its evidence and dependencies.
The output should include the proposed scope, excluded cases, required access and a way to evaluate success. That gives a future implementation a starting point and helps prevent another round of vague discovery.
A readiness assessment is useful when it reduces uncertainty. If it shows that a simpler process change solves the problem, that is a valuable finding too. The objective is a better operation, with AI used where it has a clear role.
Request an AI readiness assessment
Describe one recurring task and the information your team uses to complete it. I can help assess the process, data and integration gaps before you commit to a build.
Updated 30 September 2026.