AI Readiness Assessment PDF: Free Checklist You Can Print

This page is an AI readiness assessment you can download as a four-page PDF, with no email form in the way. It's 36 questions across six areas, starting with the work itself, and every question asks for something you can show rather than a score out of five. A 10-to-200-person business can work through it in about a week, alongside the day job.

Most of what ranks for this term wasn't written with a smaller operation in mind. Of the 21 ranking pages we could read in September 2026, none gave an employee range, 16 assessed the whole organization rather than a specific piece of work, and 13 ended in a sales pitch. Only two asked for evidence or a checkable fact. The rest used self-ratings, plain yes-or-no questions, no scoring at all, or a scoring method we couldn't see.

That matters, because in our view a self-rating tends to measure confidence rather than readiness. The questions below ask for evidence instead: a count, a file, a name, a date. If you can't produce it, the answer is "no" or "don't know," and either one tells you more than a 3 out of 5 would.

This page is the do-it-yourself companion to our guide to AI readiness assessment services, which explains what each of the six areas means and how a full assessment runs. We should be upfront that Avolis runs diagnostics for a living, which gives us a stake in how you read this. So the checklist is written to be usable without us, and it's the same one we'd want a client to have worked through before we start.

Key Takeaways

• A useful AI readiness checklist asks for evidence you can produce, such as a weekly count, an export file, or a named owner, not a 1-to-5 self-rating.
• In the New York Fed's August 2026 regional surveys, the top reason firms gave for not using AI was that their type of work didn't suit it (46% of service firms, 54% of manufacturers), which is a claim the work section can test.
• Start with the work, not the data. Data problems only matter once you know which workflow they'd affect.
• Score by sorting, not averaging: blockers, fix-alongside items, and gaps that don't matter for this workflow.
• Treat every "don't know" as a finding.

Table of Contents

• What's in this AI readiness assessment PDF?
• Why the questions ask for evidence, not ratings
• What the checklist is built to catch
• The AI readiness assessment checklist
• The AI data readiness assessment template
• How to score your answers
• AI readiness assessment tools and templates compared
• Where Avolis fits
• Frequently Asked Questions
• Continue Learning

What's in This AI Readiness Assessment PDF?

The checklist has 36 yes-or-no questions in six areas, a one-page data readiness template, and a sorting sheet for the results, with space on each to write your answer and the evidence behind it. It's written for operations-heavy businesses of 10 to 200 people, and it assumes one workflow at a time rather than the whole company, because readiness is always readiness for something specific.

The six areas follow the order we use in our own diagnostics: the work itself, data, systems, people, rules, and direction. The order is deliberate. Data appeared in every ranking checklist we read that named its areas, but data problems only matter once you know which workflow they'd affect, which is why our readiness guide puts the work first.

Here's what you'll need, and who should be involved:

• One or two candidate workflows, named plainly: "quote follow-up," "work order intake," "month-end job costing." Not "marketing" or "operations."
• The owner, and the person who does the work. The owner answers the direction questions. The office lead or estimator answers almost everything else, because they're the one who knows where the job actually stalls.
• Access to your main systems, enough to run an export and look at recent records.
• About a week. Most questions take minutes. The few that need counting or timing take a few days of noticing.

The checklist PDF holds just the questions, the data template, and the sorting sheet, laid out to print and fill in by hand. There's nothing to sign up for. If a question doesn't apply to your business, skip it and write down why, since that note is useful later.

Why the Questions Ask for Evidence, Not Ratings

Some of the clearest evidence on checklists comes from surgery. In 2009, a study in the New England Journal of Medicine found that introducing the World Health Organization's 19-item Surgical Safety Checklist at eight hospitals was associated with a drop in the death rate after surgery from 1.5% to 0.8% (Haynes and others, A Surgical Safety Checklist to Reduce Morbidity and Mortality in a Global Population, NEJM, January 2009).

Checklists spread quickly after that. Then, in 2014, a second NEJM study looked at what happened when 101 hospitals in Ontario, Canada, adopted checklists under a provincial policy encouraging every hospital to use one. Across more than 200,000 procedures, the adjusted death rate moved from 0.71% to 0.65%, and the change wasn't statistically significant (Urbach and others, Introduction of Surgical Safety Checklists in Ontario, Canada, NEJM, March 2014). Complications didn't move meaningfully either.

The two studies differ in design, size, and starting death rates, so they don't prove that one rollout was done well and the other badly. They're consistent with a simple idea, though: the benefit comes from running the checks, not from having the form. The Ontario study tracked outcomes through administrative health data and a survey of when each hospital adopted its checklist, which tells you a checklist was in place but not how each team ran it.

Our reading for AI readiness: most readiness checklists invite the Ontario outcome. A question like "rate your data quality from 1 to 5" can be answered in two seconds without looking at anything, so it tends to measure how the person answering feels about the business. A question like "in your last 20 jobs, how many have the job type filled in?" can only be answered by opening the system and counting. That's why every question below has a "counts as yes" column, and why a "yes" without the evidence is a "don't know."

There's a second reason, specific to small operations. The owner usually answers a readiness quiz, and the owner often isn't the one who does the workflow day to day. Evidence questions route each answer to whoever can actually see it, which is usually the office lead, the estimator, or the dispatcher.

What the Checklist Is Built to Catch

Most reasons businesses give for not using AI are things a checklist can test. In August 2026, the New York Fed surveyed firms in New York and northern New Jersey, and among those not using AI, 46% of service firms and 54% of manufacturers said their type of work didn't lend itself to AI (Abel and others, Businesses Are Using AI to Transform Work, Not Cut Jobs, Federal Reserve Bank of New York, September 2026).

Look at what's on the list. Skills, data privacy, whether you'd catch inaccurate output, leadership priority, and legal risk are all things you can check for a specific workflow, rather than beliefs you have to hold or drop. The top reason is the most interesting, because it's sometimes true and sometimes a guess. A crew that frames walls all day really doesn't have much for AI to do on site. The office that quotes, schedules, invoices, and chases paperwork for that crew usually does, and the work section below is built to find out which case you're in.

Cost, notably, was the lowest of the reasons the authors charted for service firms, at 8%. That fits the rest of the survey. The authors report that three-quarters of service firms and more than 90% of manufacturers characterized their AI investments as minimal to modest, and the median share of workers using AI at adopting firms was 17% in services and 7% in manufacturing. In other words, most businesses that have started did so small, with a few people and a few tools.

Citation-ready summary: In the New York Fed's August 2026 regional business surveys, 61% of service firms and 51% of manufacturers in the New York–northern New Jersey region used AI. Among firms that didn't, the most common reason was that their type of work didn't suit AI (46% and 54%), while cost was among the least cited (8% and 15%).

The survey is also useful on the people question, which the checklist asks directly. Among firms using AI, only 4% of service firms and no manufacturers reported laying off workers because of it in the past six months, while 34% of service firms and 22% of manufacturers were retraining workers. That's the pattern we see too: the change is to what people spend their day on, not how many of them there are.

The AI Readiness Assessment Checklist

Work through the six areas in order, for one named workflow at a time. For each question, answer yes only if you can produce what's in the "counts as yes" column; otherwise mark no or don't know. None of it needs special software, just your own records and the people who keep them.

Write the workflow's name at the top before you start. If you're checking two workflows, run the list twice, since a business can be ready for one and not the other.

1. The Work Itself

If questions 2 and 3 come back "don't know," stop and count for a week before going further. Everything else on the list depends on them, and in our diagnostics it's where the most useful surprises turn up.

2. Data

3. Systems

These are the short version of the six checks in our AI infrastructure readiness assessment, which explains which gaps block a first build and which don't.

4. People

5. Rules

6. Direction

Question 36 matters more than it looks. Most businesses we meet have tried something already, and the reason it stopped is usually a readiness finding in disguise: nobody owned it, it didn't connect to the system the work lived in, or the team never trusted its output. Our look at why AI projects fail covers the common patterns.

The AI Data Readiness Assessment Template

The data template turns questions 7 to 12 into one table per workflow, filled in from a sample of recent records. Among the New York Fed's manufacturers not using AI, data privacy was the second most-cited reason, at 38%, and 33% of non-adopting service firms cited it too, so the template asks where personal information sits as well as whether the records are complete.

Fill in one row for each piece of information the workflow needs. Then pull your last 20 jobs, quotes, or work orders and count how many have that field filled in correctly. Twenty is enough to see a pattern and few enough to count in an hour.

Here's how it looks filled in for a composite example: a 35-person electrical contractor checking whether it's ready to have AI draft quote follow-up emails. The company and its numbers are illustrative, not a client or a benchmark.

The table makes the decision for you. Names, emails, and amounts are ready. Job type is typed freehand, so it's filled in on only 11 of 20 and spelled five different ways. The last contact date doesn't exist anywhere a system can see, which means the one thing a follow-up needs, knowing who hasn't heard back, isn't recorded yet.

What the template usually shows: in our diagnostics, the gap is rarely "our data is a mess." It's one or two specific fields that nobody records because nobody has needed them in a system before. In the example, fixing it means adding a job-type dropdown and a status field to the CRM, which takes an afternoon. That's a very different project from the data cleanup most readiness checklists imply.

For the full version of this check, including how to judge accuracy and not just completeness, see our data readiness assessment guide.

How to Score Your Answers

Don't add the answers up. Sort them. Each "no" or "don't know" goes into one of three piles for the workflow you're checking, the same sort our infrastructure assessment applies to systems, and the piles tell you whether to start, fix something first, or pick a different workflow.

• Blocks this workflow. It has to be fixed before a build makes sense. In the electrical contractor's case, the missing last-contact date blocks follow-up emails, and a shared CRM login would too.
• Fix alongside the build. It matters, but it can be done in the first few weeks. The job-type dropdown and a one-page AI rule usually land here.
• Doesn't matter for this workflow. It's a real gap, just not this one's. A missing backup test is worth fixing anyway, but it doesn't decide whether a follow-up email draft is ready.

On the sorting sheet, write the number of each "no" or "don't know" question under the pile it belongs in:

Count your "don't know" answers separately. They're the most useful output of the whole exercise, because each one is a question nobody in the business can answer yet. In our diagnostics, they bunch up in the work section, around how often a task happens and how long it takes, which is exactly what a measured assessment exists to find out.

If most answers in the work and direction sections are "yes," and nothing sits in the blocker pile, the workflow is a reasonable candidate to build. If the blockers are all in one area, fix that area and run the list again in a month. For how a full assessment turns results like these into a go, wait, or no decision on each workflow, see the decision section of our readiness guide. How the six areas are weighted against each other is covered in our AI readiness assessment methodology.

AI Readiness Assessment Tools and Templates Compared

Most AI readiness assessment tools and templates measure the same handful of things and differ mainly in how they score. Of the 16 ranking pages we reviewed that named their dimensions, all 16 included data and 12 included governance, while only 3 tied the assessment to a specific workflow and 3 kept the checklist behind an email or sign-up form.

The public tools are worth five minutes, and our readiness guide covers the two best free self-assessments in more detail. None of these formats is wrong. They just measure different things, and a self-rated quiz can't answer a question that needs a count.

The strongest checklists we read borrow the evidence idea. BD Emerson's, for example, only gives a capability its top score when it's "documented, owned by a named person, and produces an artifact" (BD Emerson, AI Readiness Checklist). That's the right instinct, and it's the test this checklist applies to every question, scaled down for a business without an engineering team.

If you're a smaller operation wondering whether any of this is worth doing at your size, our page on AI readiness assessments for SMBs takes that question on directly.

Where Avolis Fits

This checklist covers the same six areas as our two-week diagnostic, and it's a fair test of whether you need one. The difference is the part a checklist can't do: working through each workflow with your office lead and estimator, measuring how it actually runs, and ranking every workflow in scope by impact and effort. The diagnostic is where every engagement with us starts, and you come away with that ranked list whatever you decide to do next.

If the checklist comes back clean for a workflow, you may not need anyone, and a capable tool plus a named owner could be enough. If it's full of "don't know" answers in the work section, that's the case where a measured assessment earns its cost. Whether to hire anyone, and what the options cost, is covered in our comparison of AI readiness assessment consulting firms, and the broader partner decision starts with which AI consulting company to choose.

Frequently Asked Questions

Is there a free AI readiness assessment PDF?

Yes. This page links to a four-page checklist PDF with no email form, holding 36 evidence-based questions across six areas, a data template, and a sorting sheet. Evidence-based checklists are rarer than they should be: of 21 ranking readiness pages we reviewed in September 2026, only two asked for evidence rather than opinions.

What should an AI readiness assessment checklist include?

It should cover the work itself, data, systems, people, rules, and direction, in that order, for one named workflow at a time. Each question should ask for evidence, such as a weekly count or an export file. In the New York Fed's 2026 surveys, 46% of non-adopting service firms said their work didn't suit AI, which the work questions can test.

What is an AI data readiness assessment template?

It's a table listing each piece of information a workflow needs, where it lives, who enters it, and whether it can be exported. You fill it in by counting a sample of about 20 recent records. Data privacy concerned 33% of the New York Fed's non-adopting service firms in 2026, so it should also flag personal information.

How long does an AI readiness self-assessment take?

For one workflow, about a week alongside normal work. Most of the 36 questions take minutes, and the rest need a few days of counting and timing. That time goes into producing evidence, which matters: a 2014 Ontario study found no significant drop in surgical deaths after 101 hospitals adopted checklists.

Are AI readiness assessment tools better than a checklist?

They measure different things. Free public tools and vendor quizzes take minutes and profile how you describe your business. An evidence checklist checks a workflow against things you can show. Of the 21 ranking pages we reviewed in September 2026, only two asked for evidence, so check how a tool scores before trusting its result.

Continue Learning

An AI readiness assessment PDF is only as good as the evidence behind each answer. Pick one workflow, answer the 36 questions with a count, a file, or a name, and sort what's missing into blockers, fix-alongside items, and gaps that don't matter yet.

If you do only one thing this week, count how many times that workflow happens and time it three times.

The wider assessment:

• AI readiness assessment services
• AI readiness assessment methodology
• AI readiness assessments for SMBs
• How to assess your organization's AI readiness yourself

Going deeper on one area:

• Data readiness assessment for AI initiatives
• AI infrastructure readiness assessment
• Generative AI readiness assessment

Choosing who does it:

• AI readiness assessment consulting firms
• Which AI consulting company should I choose?
• Why do AI projects fail?

---

Sources

All sources retrieved 2026-09-24.

• Jaison R. Abel, Richard Deitz, Natalia Emanuel, and Nick Montalbano, Businesses Are Using AI to Transform Work, Not Cut Jobs, Federal Reserve Bank of New York, Liberty Street Economics, September 1, 2026 (Regional Business Surveys, August 2024, 2025, and 2026; New York and northern New Jersey firms; figures from the post and its chart data workbook), retrieved 2026-09-24 — https://libertystreeteconomics.newyorkfed.org/2026/09/businesses-are-using-ai-to-transform-work-not-cut-jobs/
• Alex B. Haynes and others, for the Safe Surgery Saves Lives Study Group, A Surgical Safety Checklist to Reduce Morbidity and Mortality in a Global Population, New England Journal of Medicine 360:491–499, January 2009, retrieved 2026-09-24 — https://www.nejm.org/doi/full/10.1056/NEJMsa0810119
• David R. Urbach, Anand Govindarajan, Refik Saskin, Andrew S. Wilton, and Nancy N. Baxter, Introduction of Surgical Safety Checklists in Ontario, Canada, New England Journal of Medicine 370:1029–1038, March 2014, retrieved 2026-09-24 — https://www.nejm.org/doi/full/10.1056/NEJMsa1308261
• BD Emerson, AI Readiness Checklist, 2026 (scoring rubric; cited for its evidence rule only), retrieved 2026-09-24 — https://www.bdemerson.com/article/ai-readiness-checklist

On the New York Fed figures. The surveys cover firms in the New York–northern New Jersey region, not the whole country, and the authors note that their adoption shares sit toward the high end of the range in existing studies. Firms that used AI only as a search tool weren't counted as users. The reasons-for-not-using chart shows the share of non-adopting firms citing each reason, and firms could give more than one. The mapping of each reason to a checklist area is ours.

On the checklist studies. The WHO pilot compared consecutive patients before and after introduction at eight hospitals. The Ontario study compared three-month windows before and after adoption at 101 hospitals using administrative data. They differ in design and baseline risk, and we use them only to make a narrow point: adopting a checklist and running it aren't the same thing.

On the page review. "The 21 ranking pages we reviewed" means the distinct pages returned on 2026-09-24 for five searches: "ai readiness assessment pdf," "ai readiness assessment template," "ai readiness assessment checklist," "ai data readiness assessment template," and "ai readiness assessment tools." We excluded academic papers, national-government assessments, one off-topic page, and three pages we couldn't read. Interactive tools whose questions only appear in a live browser were counted as unclear rather than guessed at, which is why the scoring counts don't add up to a clean split. It's a snapshot of one day's results, not a market survey.

On first-party claims and examples. Statements such as "in our diagnostics" describe Avolis's own engagements. They're not independent research and not benchmarks. The electrical contractor and its record counts are an illustrative composite, not a client.

---

About Avolis Research Group

Avolis Research Group is Avolis's in-house research practice, focused on how operations-heavy small and mid-sized businesses actually adopt AI. It synthesizes primary economic research, government survey data, and results from real implementations into practical, vendor-neutral guidance.

More about Avolis and how we work · Get in touch