Best AI readiness assessment tools in 2026, and what they cannot tell you
Eight AI readiness assessments read from their own pages: what each asks, what you get back, what is sold next, and what a readiness score cannot answer.
Short answer
Use Microsoft’s AI Readiness Assessment or Cisco’s for a free first pass. Both score your whole organization on pillars such as data, people and governance. Neither names which workflow AI should take first, and no self-serve form page we read promises that. For that, study the actual work. We read eight options’ own pages on October 16, 2026.
On this page
- Options compared
- 8: seven assessments plus Nolli
- Read
- October 16, 2026, from each tool’s own page
- Completed by us
- None. We read pages and question lists.
Quick picks
- Best free first passMicrosoft AI Readiness AssessmentThe page publishes all 72 questions across seven pillars, so you can read them before you start. Skip it if you want an answer in ten minutes.
- Best for a score you can compareCisco AI Readiness AssessmentA 0 to 100 score in four levels, set against a 2025 survey of 8,039 senior leaders. Infrastructure and data carry 45% of the Index score.
- Best for a short first lookMicrosoft AI Readiness WizardTen questions, two for each of five drivers. It returns one next area to focus on, which is also its limit.
- Best for the next questionNolli (ours)Not a form. It reads the actual work to show which workflows suit AI and what each is worth. It gives no company-wide readiness score.
Nolli makes one of the products on this page. We list it like the others, with its limits. We compare from each vendor’s public pages and name our sources below. We did not run hands-on tests.
Key takeaways
- An AI readiness assessment scores your whole company on pillars such as data, people and governance, from answers that people give about it.
- Microsoft’s two forms each have one question about whether you have approved use cases, and no self-serve form we read lists a workflow among its outputs.
- Use a form to find gaps, then study the real work to decide what to build first, and prove it in shadow mode before you commit.
A home inspector checks the roof, the wiring and the pipes. She tells you whether the house can take a renovation. She does not tell you which room to renovate first, and nobody expects her to.
An AI readiness assessment is the home inspection. It asks about your data, your people, your rules and your systems, then gives you a score or a report. That is a fair thing to buy before a big AI program. The trouble starts when the score gets mistaken for the plan. We read seven assessments on their own pages, plus Nolli as the odd one out, and say below what each one asks, what you get back, and what happens next. The wider question, where AI belongs in your company, has its own page: How to find where AI belongs.
What a readiness assessment asks
Open any of them and you find the same five or six headings: strategy, data, people and skills, infrastructure, governance, and sometimes model management. Each heading is called a pillar. You answer questions about each pillar on a scale, from “nothing in place” up to “run and reviewed continuously”. The tool adds the answers up and places you on a maturity level, which is a named rung on a ladder.
AI readiness assessment
A questionnaire, or an adviser-led review, that scores how prepared an organization is to adopt AI. Our definition, drawn from the pages below. It scores the organization, from answers people give about it. It does not score a single workflow. A close relative is the AI opportunity assessment, which goes further and ranks use cases.
Two meanings of the word “ready” sit under that definition, and this post depends on telling them apart. Company readiness asks whether this company could run AI safely at all. Workflow readiness asks whether one step of one job is ready to hand over, given the judgment it takes, the data it needs and what a mistake would cost. Every tool below answers the first. The second is what you need before you decide what to build.
Seven assessments and one alternative
Microsoft AI Readiness Assessment (Microsoft Learn)
Best for a free first pass you can read before you start
A multiple choice questionnaire on Microsoft Learn. It scores seven pillars: business strategy, AI governance and security, data foundations, AI strategy and experience, organization and culture, infrastructure for AI, and model management. The page shows a length of 45 beside a clock icon, which we read as minutes. The questions sit in the page itself, and we counted 72.
Strengths
- The page says what comes out: curated, personalized guidance, with recommendations you can follow to improve your score over time.
- The full question list is readable, which no adviser-led tool here allows.
- Fine detail on data. It asks separately about ingestion, discovery, access, labeling, privacy, quality and metadata.
Limits
- Three of the 72 questions sit under business strategy, and one of those asks about use cases. Seventeen each sit under data foundations, infrastructure and model management.
- Some questions name Azure OpenAI, and others ask about “the customer’s” cloud, so it reads as written for Microsoft’s partners to run with clients. We did not confirm that.
- The page does not show how the score is worked out or what the result looks like.
Pricing No price on the page. We saw no email field.Vendor page
Cisco AI Readiness Assessment
Best for a 0 to 100 score you can set beside a published survey
An online tool tied to Cisco’s yearly AI Readiness Index. It scores six pillars: strategy, infrastructure, data, governance, talent and culture. It places you at one of four levels, from Unprepared (0 to 30) up to Fully Prepared (above 86), and its page shows the 2025 global spread: 13% Pacesetters, 36% Chasers, 48% Followers and 3% Laggards.
Strengths
- The method behind the Index is public. The 2025 report names 49 indicators and gives each pillar’s weight.
- A real benchmark: the Index sample is 8,039 senior leaders at organizations of 500 or more employees, in 30 markets, surveyed online in August 2025.
- Four named levels with stated cut-offs, so a score reads as a position and not a bare number.
Limits
- In the Index, infrastructure carries 25% of the score and data 20%. The tool page does not state its own weights.
- The page names no use case or workflow step among what it scores.
- We read the tool’s page and could not run the tool, so its length and any email request are unconfirmed.
Pricing Not published on the page.Vendor page
Microsoft AI Readiness Wizard
Best for a short first look
Ten multiple choice questions, two for each of Microsoft’s five drivers of AI value: business strategy, technology and data strategy, AI strategy and experience, organization and culture, and AI governance and security. Each question has five answers on a rising scale. Per a November 2024 Microsoft blog post, the result places you at one of five stages, from exploring to realizing. The page says you get the next best area to focus on and a few resources.
Strengths
- The answers describe situations, such as “no identified use cases”, rather than feelings, so it is harder to flatter yourself.
- It has a question about whether you hold approved and prioritized use cases for AI. Microsoft’s Learn assessment asks the same one.
- Short enough to give to three people and compare their answers.
Limits
- Ten questions see little. Data and security share two of them.
- The result is one area to focus on, not a ranked list.
- The page says the drivers come from Microsoft’s research and customer work. It shows no sample result and no scoring rules.
Pricing No price on the page.Vendor page
Eide Bailly AI Readiness Assessment
Best when you already suspect data is the problem
A ten-question assessment from an accounting and advisory firm. Its page says it gauges whether your data, processes and operating model can support AI, and that you get a snapshot of your current state plus guidance on what to prioritize.
Strengths
- It names its scope plainly: the data foundation, meaning quality, governance and accessibility.
- Ten questions, so it is quick to try.
- It does not pretend to be a verdict on the whole company, which we count in its favor.
Limits
- Data only. It does not look at people, process steps or use cases.
- The page gives no length, price or sign-in rule.
Pricing Not stated on the page.Vendor page
Augmentry AI Readiness Assessment Tool
Best as a clear example of the form that leads to a call
An online form with 40 questions across eight dimensions, about ten minutes by its page. Its sample result shows a 1 to 5 score for each dimension, an overall score and industry and company-size averages. It then offers a PDF report or a call about a paid, expert-led assessment that takes two to three weeks.
Strengths
- It tells you the length up front: 40 questions, eight dimensions, about ten minutes.
- Scores each dimension, so you see which area drags the total down.
- Its own FAQ calls the form a high-level screening tool, which is accurate.
Limits
- Contact details are step 1 of 9, before the first question.
- The benchmark figures we saw sit in a sample result with no method note, so we cannot say what sample they come from.
- The page lists no outputs that name a workflow. The next step it offers is a paid assessment.
Pricing The form is free, per its page. Contact details come first. The expert-led assessment: not published.Vendor page
RSM AI readiness assessment
Best for a mid-sized company that wants advisers to run it and name use cases
A four-week engagement, not a form. Its steps are kickoff and discovery with documents, interviews and surveys, an analysis that identifies relevant use cases, a gap analysis with return-on-investment insights and effort and cost estimates, and a prioritized roadmap. It ends in a written readiness report.
Strengths
- Of the seven assessments, its page is the only one that lists use cases among the outputs.
- It names its six pillars and says what each of the first four steps produces.
- It gathers more than one voice, through stakeholder interviews and surveys.
Limits
- The inputs are documents, interviews and surveys. No step observes the work itself.
- The page publishes no price or team, and it ends with an invitation to talk to RSM’s advisers.
- Implementation and optimization follow as RSM services, so the assessment is a front door to them.
Pricing Not published.Vendor page
PwC AI Readiness Assessment
Best for a large company that wants a scored benchmark against peers
PwC’s diagnostic asks 285+ questions, each scored 0 to 5, across 12 domains. It rolls them up to a 0 to 100 score and one of five maturity levels, and compares you with 500+ organizations in PwC’s benchmarking set. Recommendations come ranked Critical, High or Quick Win, with a 90-day, 6-month and 12-month plan.
Strengths
- It publishes how the scoring works: question count, scale, domains and maturity levels.
- Each recommendation carries impact and effort ratings and links to the domain gap behind it.
- The plan is dated, in three horizons, with milestones and dependencies.
Limits
- It scores the whole business across 12 domains that include customer experience and financial resilience. Its page lists no workflow-level output.
- Its page says clients can ask for PwC’s review of any recommendation to start an advisory engagement, and calls that a “downstream pipeline mechanism”.
- No price, sample report or duration on the page.
Pricing Not published. Contact a PwC partner.Vendor page
Nolli
Made by NolliBest when the question is which workflow to hand to AI first
Nolli is not a readiness form. It learns how work really runs from people’s work on company devices, from the email and calendar your company connects, and from the logs of the AI agents people already use. Then it shows where AI belongs, what each change is worth and how an agent does beside people on your own work. Readiness is read for each workflow, from the work itself. Nolli is ours, so weigh this entry with that in mind.
Strengths
- It starts from observed work, so nobody has to estimate hours or remember a workaround.
- It sorts steps into the four buckets (delete, plain code, agent or person) and prices each change.
- Your company decides what it may learn from. The trust page says passwords and login, checkout and billing pages stay out.
Limits
- It does not score strategy, culture, infrastructure or governance, and it gives no maturity level or benchmark against other companies. If a board wants that, use one of the forms above.
- Not a quick self-assessment. Your IT team deploys it on company devices, and it learns as people work, so it is not an answer in ten minutes.
Pricing No public price. Request a demo.Vendor page
What each one covers, and what the weights favor
Read across the grid and you see a block of full circles on the left, and two rows at the bottom that are nearly empty. The top six rows are the usual pillars. The bottom two are the questions this post is about.
| Microsoft Learn | Microsoft Wizard | Cisco | Eide Bailly | Augmentry | RSM | PwC | Nolli | |
|---|---|---|---|---|---|---|---|---|
| Strategy and leadership | fully | fully | fully | none | partly | fully | fully | none |
| Data | fully | partly | fully | fully | fully | fully | fully | none |
| People and skills | fully | fully | fully | none | fully | fully | fully | partly |
| Infrastructure | fully | none | fully | none | fully | fully | fully | none |
| Governance and risk | fully | fully | fully | partly | fully | fully | fully | none |
| Model management | fully | partly | none | none | none | fully | none | none |
| Names and ranks your workflows | partly | partly | none | none | none | fully | none | fully |
| Reads the actual work, not answers about it | none | none | none | none | none | none | none | fully |
none partly fully · Our reading of each tool’s own page, October 16, 2026. Full: the page says it plainly. Half: it says part of it, for example Microsoft’s Wizard has two data questions and one model question, and its Learn assessment has one use case question out of 72. Empty: we found nothing, which does not prove the feature is absent. Nolli’s column is mostly empty on purpose: it is not a questionnaire, and its people and skills cell is half because it shows who holds the know-how and does not assess skills. RSM’s use case cell is full because its page lists use cases among its outputs, found through interviews and surveys, not observation.
Pillars are not neutral. The weights tell you what the maker thinks matters, and Cisco publishes its weights for the Index its tool is built on.
Infrastructure and data together carry 45% of the Index score. Cisco’s own index page leads with a network finding, that only 15% of surveyed companies have networks fully ready for AI. That is a reasonable thing for a networking company to measure. It also means a company with tidy infrastructure and no idea what to automate can score well.
of organizations in Cisco’s 2025 sample were Pacesetters, the top level (above 86 of 100). The sample was 8,039 senior leaders at organizations of 500 or more employees in 30 markets, surveyed online in August 2025.
questions in Microsoft’s AI Readiness Assessment sit under three pillars: data foundations, infrastructure for AI and model management (17 each). Business strategy has 3. Our count of the question list embedded in the page on October 16, 2026.
questions in Microsoft’s AI Readiness Wizard asks whether you have approved, prioritized and socialized use cases for AI. The Learn assessment has the same question, once, among its 72.
What a readiness score cannot tell you
The forms are not wrong. They answer a narrower question than the one most buyers have. Four things are worth knowing before you act on a score.
The answers come from whoever fills it in
Microsoft’s Learn assessment is multiple choice, and Cisco’s survey is answered by senior leaders. The score is only as good as one person’s view of the company on a Tuesday. Ask “how effective is your data governance?” and a finance director and an analyst give two different answers. Both are honest. Neither is a measurement.
The maker’s product shapes the questions
Microsoft’s Learn assessment includes questions that mention Azure OpenAI by name, in its model selection and fine-tuning items. That is not a flaw for a Microsoft shop. For anyone else, a share of the 72 questions will not apply. Read the questions before you read the score.
The output is a pillar, not a workflow
You finish and learn that “data foundations” is a 2 out of 5. Which job does that stop? Invoice matching? Support triage? A score for data across the whole company cannot say, because data is ready or not ready for a particular job. The same is true of governance and skills. Two roundups we read from firms that sell in this space, Summit Trails and Certified AI Partners, make the same point, and each sells an answer to it. So do we. Weigh all three accordingly.
A high score is not the same as a good use case
Cisco’s report finds that its Pacesetters, the top band, had finalized their use cases much more often than other respondents (77%, which Cisco calls four times the global average). That is a correlation in a self-reported survey. It does not show that a high score produces use cases. It shows that companies that already know what they will build also tick more boxes.
Where an assessment sits on the path to a first pilot
If a form is the first step, what comes second? The path below is ours, built from the guide on finding where AI belongs. Read it left to right.
- 01Are we ready?A readiness assessment. Scores the company on data, people, rules and systems. Output: gaps and a maturity level.
- 02Which work should AI take?Use case discovery. Maps real workflows, who does them, how often, and what each step is worth.
- 03Which one first?Prioritization. Score each use case on measured hours, ease and risk. The sheet is in the prioritization template.
- 04Does it work on our work?A pilot in shadow mode. The agent runs beside people, sends nothing, and its output is compared with theirs.
- 05Build and roll outNow fix the gaps from step 1 that the chosen workflows need: a data source, a rule, a skill.
The order is ours. Step 1 can also come after step 2, and we would pick that when you already run pilots. The point is that gaps get fixed for the workflows you chose, not for the company at large. Nolli works at the two highlighted steps. For platforms that cover more of the path, see best AI enablement platforms in 2026.
Our advice, as a vendor with a stake in it: run a cheap form first, because ten minutes costs nothing, and treat the result as a list of questions for your IT and data owners. Do not treat it as a ranking of what to build. Then spend your real time on step 2.
How to pick
Our reading, not a measurement. Horizontal: the length each page states (10 questions, 40 questions and about ten minutes, 72 questions, 285+ questions, four weeks) and whether it needs a deployment. Cisco’s length is not stated, so its position is a guess. Vertical: how close the output gets to a named workflow and a next action, from what the page says you get back. No page we read puts a form near the top. We make Nolli, so treat its position with care. It sits high because it names workflows from observed work, and it sits right because IT deploys it and it learns over time.
Pick Microsoft’s Learn assessment if
you want a free read-through of what a mature company has in place, and someone can sit with 72 questions for an hour. Skip it if you want to know what to build first.
Pick Cisco’s if
a board wants a score set beside a published survey. Check the weights first, since infrastructure and data are 45% of the Index.
Pick the Microsoft Wizard or Eide Bailly’s if
you want ten questions to start a conversation. The Wizard is wide and thin. Eide Bailly’s is narrow and about data.
Pick RSM or PwC if
you have a budget, a few weeks, and want advisers to run it. RSM’s page lists use cases among its outputs. PwC’s is the better fit when you need a peer benchmark at a large company.
Skip any form that asks for contact details before the first question unless
you want the call. Otherwise, pick one that shows the score first.
Pick discovery from the work if
you already know the company can run AI, or you already run pilots, and the open question is which workflows to hand over. Nolli is one option. Others are in our enablement platforms list.
What a good assessment must output
Test any assessment, including ours, against what you hold in your hand at the end. A pillar score is a start. These are the things that let you act on Monday.
A useful readiness result contains
- A named list of workflows or steps, not only pillars.
- For each one, how big the job is: hours per week and how many people, and where those numbers came from.
- For each one, what blocks it, in plain words: a data source it cannot reach, a decision only a person can make, a rule it would break.
- A first pilot, how to run it in shadow mode, and what result counts as a pass.
- The evidence behind every score: who said it or what was observed, and when.
- A way to run it again when the work changes, because a score from March describes March.
Mistakes to avoid
Treating the score as the plan
A 2 on data foundations tells you where to look. It does not tell you what to build.
Do this todayDo this today: for each low pillar, name the person who owns it and ask which workflow it blocks.
One person answering for the company
Every form here is self-scored. One optimistic answerer can move a level.
Do this todayDo this today: send the same ten questions to three people at different levels. Where they disagree is your first finding.
Handing over contact details to find out what you get
Some forms show the score before they ask. Others ask first.
Do this todayDo this today: open the form and see whether an email field comes before the first question.
Fixing the whole company before choosing a workflow
Company-wide readiness work is slow and has no end. Gaps are cheaper to close for one chosen job.
Do this todayDo this today: pick one workflow and list only the gaps that stop it.
Believing a benchmark without its sample
Cisco publishes its sample, and Augmentry’s benchmark figures on its page had no method note. Ask what a number was measured on.
Where Nolli fits
Where Nolli fits, and where it does not
Nolli is AI discovery, not a readiness questionnaire. It learns how your company works from people’s work on company devices, from the email and calendar you connect, and from the logs of the AI agents your people use. Then it shows which workflows suit AI, what each change is worth, and how an agent performs beside your people on your own work. It does not score your strategy, culture, infrastructure or governance, and it gives no maturity level or benchmark against other companies, so use a form above if you need that. Your IT team deploys it, and your company decides what it may learn from. No public price. Request a demo.
Request a demoQuestions people ask
What is an AI readiness assessment?
It is a questionnaire, or an adviser-led review, that scores how prepared an organization is to adopt AI, usually across pillars such as strategy, data, people, infrastructure and governance. It scores the organization from answers people give about it, not a single workflow.
Are AI readiness assessments free?
Some are. The forms we read from Microsoft, Cisco and Eide Bailly show no price on their pages, and Augmentry says its form is free but asks for contact details first. The adviser-led versions from RSM and PwC publish no price.
How long does an AI readiness assessment take?
It varies by tool. Microsoft’s Wizard has ten questions, Augmentry lists about ten minutes for 40 questions, Microsoft Learn lists a length of 45 for 72 questions, and RSM runs four weeks. The Cisco, Eide Bailly and PwC pages we read do not state a length.
Does an AI readiness score tell me which AI use cases to build?
No. A readiness score describes the organization, such as the state of its data and governance. It does not rank workflows. To decide what to build first, map the real work and score each use case, then test the best one in shadow mode.
What should I do after an AI readiness assessment?
Turn each low pillar into a question for its owner, list your workflows, and score them on hours, ease and risk. Our prioritization template has a sheet for that. Close only the gaps that stop your chosen workflows.
Is Nolli an AI readiness assessment?
No. Nolli studies the actual work to show which workflows suit AI and how an agent performs on them. It gives no company-wide score or benchmark, so it complements a readiness form rather than replacing one.
Sources
- “AI Readiness Assessment.” Microsoft Learn Assessments, accessed October 16, 2026.
- “AI Readiness Wizard.” Microsoft Adoption, accessed October 16, 2026.
- Carissa Eicholz. “A strategic approach to assessing your AI readiness.” The Microsoft Cloud Blog, November 6, 2024.
- “Cisco AI Readiness Assessment.” Cisco, accessed October 16, 2026.
- “Cisco AI Readiness Index.” Cisco, accessed October 16, 2026.
- “Cisco AI Readiness Index 2025: Realizing the Value of AI.” Cisco, October 2025.
- “AI Readiness Assessment.” Eide Bailly, accessed October 16, 2026.
- “AI Readiness Assessment Tool.” Augmentry, accessed October 16, 2026.
- “AI readiness assessment.” RSM US, accessed October 16, 2026.
- “AI Readiness Assessment.” PwC US, accessed October 16, 2026.
- “AI-Ready Score: Free 7-Minute AI Assessment.” Main & Machine, accessed October 16, 2026.
- “The Readiness Check.” Rebel Minds AI, accessed October 16, 2026.
- “7 Best AI Readiness Assessment Tools & Frameworks Compared.” Seampoint, accessed October 16, 2026.
- Wendy Kinney. “Best AI Readiness Assessment Tools: An Honest 2026 Buyer’s Guide.” Summit Trails, July 16, 2026, updated September 29, 2026.
- Brenden Parker. “AI Readiness Tools vs Services.” Certified AI Partners, August 27, 2026, updated October 4, 2026.
Plain-text copy: /blog/best-ai-readiness-assessment-tools-2026.md