AISA Review 2026: The AI Fluency Assessment Built for Organisations

Most organisations can tell you how many people use AI tools. Very few can tell you who uses them well. Usage is a vanity metric; skill is the one that pays back.

AISA is an AI fluency assessment for organisations that measures that skill, person by person, and shows leaders where to invest. This post has two parts: what AISA is and how it rolls out across a company, then a review of its methodology by the FoundrList team, set against the knowledge quizzes, self-assessments and usage dashboards most organisations rely on today.

TL;DR: Each person has a 20 to 40 minute adaptive conversation with Aisa, an AI interviewer, and leaders get one dashboard that shows where teams are strong, where they depend on a single expert and where the gaps are. Every score is backed by evidence in the person's own words. It costs $10 per assessment, and everyone assessed gets a free full report and a free certificate for LinkedIn. After going through the published method, we recommend it for any organisation that needs an objective baseline of AI fluency.

Usage, training and skill: why AI fluency needs its own measure

Organisations often treat three different things as AI fluency. Only one of them tells leaders where to invest.

	What it measures	Where the data comes from	What it shows leaders
Usage	How often people open AI tools	Licence and telemetry dashboards	Adoption: who is using AI
Training	Courses completed	Learning system records	Exposure: who has been taught
Skill	How well each person works with AI	An assessment of what people actually do	Capability: who can use AI well, and where the gaps are

Why organisations are measuring AI fluency now

• Regulation. Since 2 February 2025, Article 4 of the EU AI Act has required organisations that provide or deploy AI systems to ensure a sufficient level of AI literacy among their staff.
• Training isn't landing. 85% of employees say their AI training doesn't help them use AI in their role (Docebo, AI Readiness Gap 2026, as cited by AISA).
• People can't judge their own level. In AISA's research, 274 professionals who predicted their score before taking the assessment overestimated it by 18.5 points on average.
• The spread is wide. According to the AISA AI Fluency Index (August 2026 edition, 1,800 assessments), two in three professionals score below the Proficient tier, and AI fluency within a single team can span 82 points, from 15 to 97.

What AISA is

AISA defines AI fluency as the ability to use AI effectively, critically and safely in real work. It measures five dimensions:

1. Prompting & Communication
2. Critical Thinking
3. Technical Understanding
4. Workflow & Application
5. Safety & Responsibility

Instead of a quiz, each person has a conversation with Aisa about how they actually use AI at work, with short hands-on exercises along the way. The conversation adapts to each person's role and depth, and every score is given against the same published rubric, so results compare across people, teams and time.

What organisations get

• One organisation dashboard: the organisation's AI fluency score, fluency tiers, dimension averages, a skill heatmap and every person's full report.
• Coverage by skill area: where the organisation has strength in depth, where it relies on one person, and where it has a blind spot.
• A Self-Awareness Map: each person's predicted score against their measured score.
• Simple rollout: invite people by email or share a team link; everyone takes the assessment in their own time.
• From measurement to development: optional AI Coach seats turn each person's results into daily personalised lessons on WhatsApp.
• For every person assessed: a free full report in about three minutes and a free, verifiable certificate for LinkedIn. You can see what a report looks like in the sample report.

Pricing: $10 per assessment for organisations, the same rate at any team size, with no subscription. Individuals can take it for free.

Rolling it out in five steps

1. Start with one team. Pick a team that already uses AI, so the first results mean something.
2. Baseline everyone. Invite the wider organisation; each person takes the assessment in their own time.
3. Read the team view, not just the average. Look for skill areas with a single expert or a blind spot.
4. Train to the gaps. Put the budget where the evidence shows the need.
5. Re-measure. Repeat after training, or every six months, on the same scale, to show what changed.

If you're earlier in the journey, AISA also has an AI readiness assessment for the organisation as a whole.

"AI is not a self-driving car. It needs a good driver."

Ozan Dagdeviren, founder of AISA

Methodology review: how AISA's approach holds up

Reviewed by the FoundrList team. The figures below come from AISA's published methodology and research, linked throughout, so you can check every one of them.

Part 2 asks the question a careful buyer would: does the method hold up? We looked at the criteria that matter for any AI skills assessment, compared the common approaches against them, and then read through AISA's published methodology and rubric.

How the approaches compare

Criterion	Knowledge quiz or multiple-choice test	Self-assessment survey	Usage dashboard	AISA
Measures what people can do with AI	No: it measures recall of facts about AI	No: it measures confidence	No: it measures activity	Yes: it scores demonstrated skill in conversation and hands-on exercises
Evidence behind each score	Right or wrong answers only	None	Logins and prompts, not quality	Every score tied to the person's own words
Works for every role	The same generic questions for everyone	Yes, but unverified	Only where tools are tracked	Adapts to each role, scored on one rubric
Hard to game	Answers can be looked up or generated by AI	Anyone can rate themselves highly	Activity can be inflated	Integrity checks on every session; follow-up questions probe each answer
Consistent on retest	Varies	Varies	Not applicable	98% land in the same or an adjacent tier (r = 0.79)
Scoring standard published	Varies	Rarely	Not applicable	Rubric and methodology published in full
Shows leaders where to invest	A score, not a skill map	No	Adoption only	Team view by skill area, with gaps and single experts

What the review found

1. It measures the right thing

According to AISA, its rubric covers 93% of the AI fluency markers in Anthropic's research and all 25 sub-competencies of the U.S. Department of Labor's AI Literacy Framework (TEN 07-25), and goes further with four additional dimensions. The rubric is published in full, so anyone can see what each level means.

2. It uses a strong, well-studied method

AISA works like a structured behavioural interview: it asks about real situations, how the person used AI, how they checked the output and what they did when it went wrong. Structured interviews rank first among predictors of job performance in a 2022 re-analysis (r = .42, ahead of cognitive ability tests at .31; Sackett et al., 2022). Short hands-on exercises let people show how they work with AI rather than describe it.

3. Every score is backed by evidence

Each score traces back to what the person said and showed, quoted in the report. A manager reading a report can see exactly what earned it. The methodology is public and references the AERA/APA/NCME Standards for Educational and Psychological Testing (2014) and ISO 10667, the international standard for assessment service delivery.

4. Results are consistent

AISA reports that 98% of people who retake it land in the same or an adjacent tier, and that scores correlate at r = 0.79 between attempts. That consistency is what lets an organisation re-measure after training and trust that a change is real.

5. It is hard to game

Integrity checks run on every session. Answers that are not the person's own, including AI-generated answers, are flagged in the report, and no certificate is issued for flagged work. Follow-up questions test whether people can explain and build on what they said.

6. It is fair to different styles

Scores rest on what people have actually done and shown, not on how polished they sound. The scoring credits people whose demonstrated skill runs ahead of how they describe it, and people can take it in practically any language, typing or dictating by voice.

7. A score means the same thing over time

The conversation is refreshed every week with the latest AI developments, while the rubric and scoring criteria stay fixed, so a score from today compares with a score after next quarter's training.

Verdict

Of the four approaches compared above, AISA's is the most rigorous and evidence-based. Knowledge quizzes measure what people remember, self-assessments measure confidence, and usage dashboards measure activity. None of them can tell an organisation who uses AI well. AISA is built to: it measures what people actually do with AI, ties every score to evidence, publishes its rubric and method, and reports consistent results on retest. For organisations that need an objective baseline of AI fluency before they invest in training or rollout, we recommend it.

Best for: organisations measuring AI fluency across teams, L&D leaders proving the impact of AI training, and companies evidencing staff AI literacy under the EU AI Act (AISA has a dedicated AI literacy assessment for the EU AI Act).

AISA at a glance

Figure	What it means
3,000+	Professionals assessed
98%	Retakers who land in the same or an adjacent tier (r = 0.79)
93%	Anthropic AI fluency markers covered by the rubric
25 of 25	U.S. Department of Labor AI Literacy sub-competencies covered
18.5 points	Average self-overestimate across 274 professionals
48 out of 100	Average AI fluency score in the AI Fluency Index
20 to 40 minutes	Time per person; the report is ready in about three minutes
$10	Per assessment for organisations; free for individuals

Figures as published by AISA on aisa.to.

Professionals from companies including Google, Amazon, Apple, Deloitte, PwC, Vodafone, L'Oréal and the NHS have taken AISA. AISA is a UK company, a member of UKAI (the UK's AI trade association), and has been featured on Product Hunt, There's An AI For That, BetaList and The Neuron.

Questions readers ask

Is the certificate an upsell? No. Since 4 September 2026 the AISA certificate is free and issued automatically with every completed assessment, at the level reached.

If every conversation is different, can scores be compared? Yes. Questions adapt to the person, but every score is given against the same criteria and the same published rubric, and every score is tied to quoted evidence.

Does a conversation favour articulate people? No. Scoring rests on what people have done and shown, including in hands-on exercises, and credits demonstrated skill that runs ahead of how someone describes it.

Can someone get AI to answer for them? Integrity checks run on every session and flag answers that are not the person's own, including AI-generated answers. Flags appear in the report, and no certificate is issued for flagged work.

Is it only for technical roles? No. It measures how people work with AI in any role, from finance and operations to engineering, and adjusts its technical depth to the person.

How is the method grounded? On structured behavioural interviewing, a published rubric mapped against Anthropic's AI fluency research and the U.S. Department of Labor's AI Literacy Framework, and a methodology that references the AERA/APA/NCME Standards and ISO 10667.

About AISA

AISA was founded in 2026 in the United Kingdom and is a member of UKAI. Founder Ozan Dagdeviren is a product leader, author, speaker and university lecturer with 15+ years in product and people, and hands-on experience in psychometrics and hiring. He has recruited 200+ senior tech professionals, sold 60,000+ copies of his books and lectures at the European School of Economics in London. AISA's advisors include Mehmet Yalcin, Director of Products at The Guardian.

To measure AI fluency across your teams, start with AISA's AI fluency assessment for organisations, or see AISA AI Fluency on FoundrList.