“Is our business ready for AI?” is too broad to be useful. A company can be ready to automate meeting follow-ups and completely unready to automate pricing, hiring, payments, or customer decisions.

Readiness is better judged at the workflow level. Pick one task, score it, identify the gaps, and decide whether the next step should be a pilot, more preparation, or no automation at all.

If you have not chosen the task yet, start with our guide to what a small business should automate first. Once you have a candidate, use the scorecard below.

The rule for this checklist

Score the workflow that exists today—not the version you hope to have after buying software. Readiness problems hidden during planning usually reappear as expensive implementation problems later.

The 20-point AI Readiness Score

Give the workflow 0, 1, or 2 points for each of the ten areas below. Use 0 when the answer is “no,” 1 when the answer is “partly,” and 2 when the answer is clearly “yes.” The total is a decision aid, not a certification.

How to interpret the score

  • 16–20 points: the workflow may be ready for a small, controlled pilot if there are no red flags.
  • 10–15 points: fix the weakest areas before buying or integrating a new tool.
  • 0–9 points: the process probably needs operational cleanup before AI will help.

A high total does not override a serious risk. A workflow touching sensitive data or consequential decisions can still be a poor first project even if every other part is well organized.

1. The problem is specific

2 points: you can describe the problem in one sentence, including who does the work, what takes time, and how often it happens.

1 point: everyone recognizes the pain, but the task or scope is still vague.

0 points: the goal is simply “use AI,” “save time,” or “modernize the business.”

Better problem statement

Instead of “we need AI for customer service,” write: “Our office manager spends about four hours each week turning technician notes into customer-ready status updates.”

2. The current process can be explained

2 points: the person doing the work can explain the normal steps, common exceptions, and what a correct result looks like.

1 point: most of the process is understood, but important exceptions live in people's heads.

0 points: the task is performed differently every time or nobody agrees on the correct process.

If the score is 0, document the work before automating it. A tool can make an inconsistent process run faster without making it better.

3. You have examples of good and bad outcomes

2 points: you have real examples showing normal cases, edge cases, and unacceptable output.

1 point: a few examples exist, but they are not organized or representative.

0 points: the team would be testing a new tool without a reference set.

Examples give you something objective to test against. For an email workflow, save strong past replies. For document drafting, pair source notes with approved finished documents. For classification, include examples that are easy to confuse.

4. One person owns the pilot

2 points: one named person can answer questions, approve changes, review results, and decide whether the pilot continues.

1 point: ownership is shared but someone can probably take the lead.

0 points: “the team” owns it, or everyone assumes somebody else will maintain it.

The owner does not need to be technical. In many cases, the best owner is the person who knows the work well enough to notice a polished-looking answer that is actually wrong.

5. You recorded a baseline before changing anything

2 points: you know the current time, error rate, delay, volume, or another meaningful measure.

1 point: you have a rough estimate but no consistent measurement.

0 points: success will be judged by whether the new tool “feels faster.”

Your baseline can be simple. Time ten examples. Count missing fields for one week. Record how many follow-ups are late. A small sample measured before the pilot is more useful than a complicated dashboard created afterward.

6. The consequence of a wrong answer is understood

2 points: mistakes are easy to detect and correct before they cause harm.

1 point: mistakes matter, but a defined review step can usually catch them.

0 points: an incorrect output could directly affect money, legal obligations, health, employment, safety, or another consequential decision.

A 0 here is a red flag for a first project. It does not mean the workflow can never use AI. It means the controls and expertise required are probably beyond a low-risk experiment.

7. The data is appropriate for the tool

2 points: you know what information enters the workflow and you have confirmed the chosen tool is approved for that information.

1 point: the data is mostly understood but retention, training use, permissions, or vendor terms still need review.

0 points: people are likely to paste customer, employee, financial, contractual, or other sensitive data into an unreviewed tool.

Before testing, list the data categories the workflow touches. Do not treat “the AI needs it” as a reason to expose information that your business would not otherwise send to an outside service.

8. Human review is a real step, not a slogan

2 points: a named reviewer knows what to check, sees the source information, and can stop or correct the workflow.

1 point: someone reviews the output, but the criteria or responsibility are informal.

0 points: output will be sent, published, charged, or acted on automatically even though reliability is unproven.

Define the review checklist

For a customer update, a reviewer might check names, dates, prices, commitments, missing information, and tone. “A human looks at it” is not enough if nobody knows what that person is supposed to catch.

9. The pilot has boundaries

2 points: the test has a start and end date, limited users or cases, a maximum spend, and a clear way to stop.

1 point: the team plans to “try it for a while” but has not defined all the limits.

0 points: the new tool will immediately become part of normal operations for everyone.

Boundaries make a pilot reversible. That matters because the purpose of a pilot is to learn—not to justify a purchase you already made.

10. The people doing the work understand the goal

2 points: the people affected know what repetitive work the pilot is meant to reduce and what still requires human judgment.

1 point: they know a new tool is coming but have not helped shape the workflow.

0 points: the automation is being imposed on a process without input from the people who understand its exceptions.

The employee closest to the task often knows the details a product demo will miss: unusual customer requests, workarounds, incomplete inputs, and the shortcuts that keep the process moving on busy days.

Stop and fix these red flags before piloting

The score is useful, but some issues deserve a hard stop. If any item below is true, resolve it before connecting a new AI tool to live work.

Red-flag checklist

  • Nobody can explain what a correct output looks like.
  • The workflow uses sensitive information and nobody has reviewed how the vendor handles it.
  • The first version would make consequential decisions without meaningful human review.
  • No one owns the workflow after setup.
  • There is no way to measure the current process.
  • The team cannot pause the automation without disrupting normal work.
  • The project requires changing several systems, departments, and processes at once.

Turn the score into a one-page pilot brief

If the workflow looks ready, write the pilot down before shopping for software. A one-page brief prevents the tool from defining the project for you.

Copy this pilot brief

Problem: What repetitive task are we improving?
Current baseline: How long does it take or what goes wrong today?
Input: What information enters the workflow?
Desired output: What should the system produce?
Human review: Who checks it, and what do they check?
Success measure: What number must improve?
Limits: Which users or cases are included, for how long, and at what maximum cost?
Stop condition: What result would make us end the test?

That brief becomes your buying criteria. Instead of asking whether a product has impressive AI features, you can ask whether it performs this specific job safely and measurably.

Worked example: scoring a meeting follow-up workflow

Consider an illustrative small business where a manager spends time after customer calls writing summaries and assigning action items. The company wants AI to create the first draft from approved notes or a transcript.

Total: 17/20. That does not mean “go buy a tool.” It means the workflow appears suitable for a small pilot once two gaps are closed: measure the baseline and verify how the selected tool handles the conversation data.

What a four-week pilot could look like

A useful pilot does not need a long implementation plan. For a low-risk workflow, a short cycle keeps the team focused on evidence.

Four-week pilot outline

  • Week 1 — Baseline: measure the current process and assemble test examples.
  • Week 2 — Controlled test: run the workflow on historical or low-risk examples and fix obvious failure modes.
  • Week 3 — Limited live use: use it with a small number of real cases while keeping full human review.
  • Week 4 — Decision: compare time, quality, errors, and cost with the baseline; continue, change, or stop.

The decision at the end does not have to be “roll it out” or “AI failed.” You may learn that a standard automation works better, that the process needs cleanup, or that the task does not happen often enough to justify another subscription.

If your score is low, fix operations before technology

A low readiness score is useful information. If the process is unclear, document it. If examples are missing, collect them. If success cannot be measured, record a baseline. If data handling is uncertain, review the tool before using live information.

Those changes have value even if you never add AI. They make the work easier to train, audit, improve, and eventually automate with whatever technology is appropriate.

Readiness is not about being “advanced”

A spreadsheet-based business can be ready for a focused AI pilot if the workflow is clear and controlled. A company with expensive software can be unready if nobody owns the process or knows how to judge the output.

Your next decision

If the workflow scores well and has no red flags, the next step is not a giant implementation. Use the pilot brief above to compare a small number of options against the same real examples.

Our guide to choosing your first AI tool without wasting money shows how to run that comparison without letting a product demo define your requirements.