What is an AI readiness assessment?

It is a structured look at whether a business can get value from AI now, and what stands in the way. A useful one covers five areas: whether there is a specific, valuable job for AI to do, whether the data it needs is available and clean enough, whether your systems can be connected, whether people and processes are ready, and whether risk is under control.

What are the five pillars of AI readiness?

Different frameworks name them differently, but they cover the same ground: a clear use case with measurable value, data that is accessible and good enough, systems that can be integrated, people and processes ready to work with the result, and governance, meaning security, privacy and clear responsibility. Weakness in any one can stall a project, and data and ownership are the most common.

How do I know if my business is ready for AI?

You are ready for a first project if you can name one repetitive job that takes at least ten hours a week, the information needed to do it lives in systems you can access, one person owns the process and wants it improved, and you have decided which AI tools and data are allowed. You do not need perfect data or an AI strategy to start.

What is the most common reason AI projects fail?

In small and mid-sized businesses the usual causes are ordinary: no specific problem was chosen, the data turned out to be scattered or contradictory, nobody owned the result after launch, or the process being automated was itself unclear. The AI model is rarely the weak point. That is why most of this checklist is about your business and little of it is about AI.

How long does an AI readiness assessment take?

This self-assessment takes about thirty minutes if the right people are in the room. A professional assessment that also maps your processes, checks your systems and data, estimates the value of each opportunity and prices the work typically takes one to three weeks for a small or mid-sized business.

— AI AutomationSep 21, 2026

AI readiness checklist: 25 questions to answer before you spend money on AI

A practical AI readiness checklist for small and mid-sized businesses. Twenty-five yes-or-no questions across five areas: the use case, your data, your systems, your people and process, and risk. Includes a simple score, what each result means, and how to fix the gaps that stop most AI projects.

— TL;DR

AI readiness comes down to five things: a specific job worth doing, data the AI can reach, systems it can connect to, people who own the process, and sensible rules. Score yourself on 25 questions. A high score means build now. A middling score means run a small pilot while you close the gaps. A low score means fix the foundations first.

AI readiness comes down to five things: a specific job worth doing, data the AI can reach, systems it can connect to, people who own the process, and sensible rules about risk. This checklist turns those into 25 questions you can answer in about half an hour. It is written for small and mid-sized businesses, so it asks about your operations, not about data lakes and centers of excellence.

Most AI projects that disappoint do so for ordinary reasons. No specific problem was chosen. The data was scattered. Nobody owned the result. The checklist is built around those reasons.

#How to use it

Answer each question yes (2 points), partly (1 point) or no (0 points). Be strict. Do it with the person who runs the process you have in mind, not only with whoever is most enthusiastic about AI.

Have one candidate job in mind as you go, for example "answering order-status emails" or "entering supplier invoices". Readiness is always readiness for something.

#1. The use case

#	Question	Score
1	Can you name one specific, repetitive job you want AI to do, in a single sentence?
2	Do you know roughly how many hours a week that job takes today?
3	Does it take at least ten hours a week across the team?
4	Could you write the rules for doing it on a page or two?
5	Can you say how you would know it is working, in numbers?

Why it matters. "We should use AI" is not a project. "Reply to order-status emails, which take 15 hours a week, within five minutes and with the correct tracking status" is. The second one can be priced, built and measured.

#2. Your data and knowledge

#	Question	Score
6	Is the information needed to do the job written down somewhere, not only in people's heads?
7	Is it stored digitally, in a small number of known places?
8	Is it reasonably current and free of contradictions?
9	Do you have past examples of the job done well, such as answered emails or processed invoices?
10	Do you know which of this information is personal, confidential or regulated?

Why it matters. An AI system answers from what you give it. Three versions of the returns policy will produce three different answers. Past examples matter because they become the test set that proves the system works.

#3. Your systems

#	Question	Score
11	Do the systems involved have an API or built-in integrations?
12	Do you have administrator access to them, or know who does?
13	Is there a single place where a customer, order or job record is considered correct?
14	Can you create a limited-permission account for an AI system to use?
15	Do the systems record who changed what?

Why it matters. Connecting systems is usually most of the work and most of the cost. A system with no API does not rule a project out, but it changes the price, so it is better to know now.

#4. Your people and process

#	Question	Score
16	Does one named person own this process and want it improved?
17	Is the process itself clear and stable, or does everyone do it differently?
18	Will the people who do the job today be involved in designing and testing the change?
19	Is someone available for a few hours a week, for a few weeks, to answer questions and test?
20	Is there someone who will look after the system after launch?

Why it matters. Automating an unclear process produces unclear results faster. And a system that nobody owns after launch slowly goes wrong, because your business keeps changing and it does not.

#5. Risk and governance

#	Question	Score
21	Have you decided which AI tools staff may use, and with what information?
22	Do you know what a mistake in this job would cost, and who would notice?
23	Are you clear which steps need a person to approve them?
24	Do you know which privacy, industry or contract rules apply to the data involved?
25	Is there a budget, even an approximate one, and someone who can approve it?

Why it matters. None of this needs to be elaborate. A one-page policy covers question 21, and our AI acceptable use policy template is free to copy. Questions 22 and 23 decide how much testing and review the system needs, which is a large part of its cost.

#What your score means

Add up the points. The maximum is 50.

Score	What it means	What to do
40 – 50	You are ready to build.	Get the job scoped and priced. Most of the risk has already been dealt with.
25 – 39	You are ready for a pilot.	Start small, with a person reviewing the output, and close the gaps alongside it. Look at which section scored lowest.
Below 25	Foundations first.	This is normal, and usually cheap to fix. Work through the lowest section before paying for any build.

A low score in one section matters more than the total. A business that scores 10 out of 10 on four sections and 2 on data is not ready, whatever the total says.

#Fixing the common gaps

No clear use case (section 1). For one week, have the team note every repetitive task and the minutes it takes. Rank by total hours. The top of that list is your use case. The AI automation ROI calculator will tell you what each is worth.

Scattered or contradictory information (section 2). Pick the one job. Gather only the documents that job needs. Settle the contradictions and write down one correct answer for each common question. This is a few days of work, and it improves how your team works whether or not you ever use AI.

Systems that do not connect (section 3). List the systems the job touches, and check each one's integrations page. Where there is no API, ask the supplier what is possible, and consider whether an export, a shared inbox or a scheduled report could bridge the gap.

Nobody owns it (section 4). Do not start until someone does. It does not need to be a technical person. It needs to be the person who cares whether the job is done well.

No rules (section 5). Adopt a one-page AI policy, decide which actions always need a person's approval, and write down the rules that apply to your data. An hour or two.

#What this checklist does not cover

It does not tell you which opportunity in your business is worth the most, what it would cost to build, or whether an existing product already does it. Those need someone to look at your actual processes and systems.

That is what our AI Opportunity Audit is for. It takes one week and costs $1,500. We speak with the people doing the work, review your systems and data, and give you a written report: opportunities ranked by value, with the hours saved, cost and payback estimated for each, a readiness check, and a fixed-price quote for the top recommendation. If you would like to talk first, book a free 20-minute call.

For what typical projects cost, see AI automation consulting costs and AI development costs by project type.

Common questions.

What is an AI readiness assessment?
It is a structured look at whether a business can get value from AI now, and what stands in the way. A useful one covers five areas: whether there is a specific, valuable job for AI to do, whether the data it needs is available and clean enough, whether your systems can be connected, whether people and processes are ready, and whether risk is under control.
What are the five pillars of AI readiness?
Different frameworks name them differently, but they cover the same ground: a clear use case with measurable value, data that is accessible and good enough, systems that can be integrated, people and processes ready to work with the result, and governance, meaning security, privacy and clear responsibility. Weakness in any one can stall a project, and data and ownership are the most common.
How do I know if my business is ready for AI?
You are ready for a first project if you can name one repetitive job that takes at least ten hours a week, the information needed to do it lives in systems you can access, one person owns the process and wants it improved, and you have decided which AI tools and data are allowed. You do not need perfect data or an AI strategy to start.
What is the most common reason AI projects fail?
In small and mid-sized businesses the usual causes are ordinary: no specific problem was chosen, the data turned out to be scattered or contradictory, nobody owned the result after launch, or the process being automated was itself unclear. The AI model is rarely the weak point. That is why most of this checklist is about your business and little of it is about AI.
How long does an AI readiness assessment take?
This self-assessment takes about thirty minutes if the right people are in the room. A professional assessment that also maps your processes, checks your systems and data, estimates the value of each opportunity and prices the work typically takes one to three weeks for a small or mid-sized business.

— Want help with this?

AI automation that gives your team hours back ↗

We find the repetitive work that costs your team the most time, then build the AI workflow automation that takes it over. One workflow at a time, each measured against the hours it saves.

See the service

— Keep reading