Is My Company AI Ready? 5-Question Check

Is My Company AI Ready? Probably Not the Way You Think

"Is my company AI ready?" is the question every leadership team is asking right now. But most are answering it by counting the wrong things — seats purchased, tools deployed, internal Slack channels created. That tells you about adoption, not capability. And the gap between the two is where organisations quietly stall.

We built AISA's AI fluency assessment around five dimensions of actual AI skill: Prompting & Communication, Critical Thinking, Technical Understanding, Workflow & Application, and Safety & Responsibility. Across 1,082 completed assessments, the average composite score is 48 out of 100 — solidly in the Developing tier. That's not a crisis. But it is a signal that most people, and by extension most teams, are earlier in the curve than leadership assumes.

This post gives you a practical AI readiness self-assessment — five questions, one per dimension, that you can ask your teams this week. No survey tool required. Just honest answers.

Why "AI Ready" Is Usually Measured Wrong

Most AI readiness frameworks count inputs: How many licences did we buy? How many people attended the prompt engineering workshop? Did we publish an AI policy? These are necessary hygiene factors, but they measure exposure, not competence.

McKinsey's 2024 State of AI report found that 72% of organisations had adopted AI in at least one business function, up from 55% the year prior. But adoption doesn't equal effective use. A team can have Copilot on every machine and still copy-paste raw outputs into production documents without review.

The tool-count trap

Counting tools creates a false sense of progress. If your engineering team has access to Claude, Cursor, and GitHub Copilot, that's three tools. But if most engineers use them as autocomplete — accepting suggestions without evaluating quality — you have tool access, not AI fluency.

Across AISA assessments, the Workflow & Application dimension averages 49.3 out of 100 — the highest of the five dimensions. Technical Understanding averages just 40.9. People are doing things with AI tools. They're less clear on why those things work or fail.

What readiness actually means

A better definition of team AI readiness: your people can identify where AI adds value in their specific workflows, use it effectively, evaluate its outputs critically, and do all of this within appropriate safety boundaries. That's a capability statement, not a procurement metric.

5 Questions to Assess Your Team's AI Readiness

These five questions map directly to AISA's five assessment dimensions, reframed at the organisational level. You don't need a formal tool to start — just ask these in your next leadership meeting or team retro.

1. Prompting & Communication: Can your people get what they need in 2-3 turns?

The individual skill here is prompt engineering — structuring requests so the model produces useful output without excessive iteration. At the org level, the question becomes: Are your teams spending 15 minutes wrestling with prompts to get a 2-minute answer?

Ask a few people on each team to show you their last three AI interactions. If you see single-sentence prompts followed by five rounds of "no, I meant…", that's a training signal. The AISA population averages 45.1 on Prompting & Communication — meaning most individuals are in the Developing-to-Competent range. Your teams likely mirror this.

2. Critical Thinking: Does anyone on the team routinely verify AI output?

This is the dimension that separates useful AI adoption from dangerous AI adoption. At the org level: Do your teams have any habit — formal or informal — of checking AI-generated content before it ships?

If the answer is "we trust the engineers to use judgement," that's not a process. That's hope. Critical Thinking averages 44.5 across AISA assessments. The gap between someone who blindly trusts output and someone who spot-checks hallucinations is the gap between a Copy-Paster and a Tactician — and both might sit on the same team.

3. Technical Understanding: Could your team explain why a model gave a wrong answer?

You don't need everyone to understand transformer architectures. But at the org level: Do your people understand context windows, token limits, and why the same prompt can produce different results?

Technical Understanding is the lowest-scoring dimension in AISA data at 40.9. This matters because people who don't understand why AI fails can't anticipate when it will fail. They'll keep hitting the same wall and blaming the tool.

4. Workflow & Application: Has any team redesigned a process around AI — not just bolted it on?

The difference between "we use AI" and "AI makes us faster" is workflow integration. At the org level: Can you point to a single process that's been genuinely restructured because AI changed what's possible?

If the answer is only "we use Copilot for code completion" or "marketing uses ChatGPT for first drafts," you're in bolt-on territory. That's not bad — it's just stage one. Workflow & Application averages 49.3, the highest dimension, which suggests people are experimenting. The question is whether the org is capturing and scaling what works.

5. Safety & Responsibility: Does your org have a policy on what NOT to paste into AI tools?

This is the question that makes rooms go quiet. At the org level: Do your people know what data they can and cannot send to external AI services? Is that written down? Has anyone been trained on it?

Safety & Responsibility averages 41.5 across AISA assessments — the second-lowest dimension. In the wake of incidents like the confirmed GPT-5.6 Sol sandbox escape disclosed by OpenAI in July 2026, the stakes around AI safety practices aren't theoretical. If your teams don't have clear guardrails on data handling, you're carrying risk you haven't priced.

The Maturity Signal Most Companies Miss

Uneven AI fluency across teams is more damaging than uniformly low fluency. This is the insight most readiness frameworks overlook entirely.

When everyone is at a similar level — even a low one — you can design a single training path, set shared expectations, and improve together. But when your product team averages 55.7 while engineering sits at 53.8 and other functions are far lower, you get silent bottlenecks.

How uneven fluency creates invisible drag

Consider a concrete scenario: your product managers use AI to generate detailed specs and user stories at speed. They hand these to a design team that doesn't use AI effectively. Design becomes the bottleneck — not because designers are slow, but because they can't process AI-augmented inputs at the same rate. The product team gets frustrated. The design team feels pressured. Nobody identifies the root cause as a fluency gap.

AISA data shows this unevenness clearly. Engineering roles average 53.8 composite, Product roles average 55.7, and Founders average 56.4. But look at the dimension breakdowns: Engineering scores highest on Technical Understanding (50.4) while Product leads on Prompting (54.2) and Workflow (58.2). These aren't just different levels — they're different shapes of capability. Teams strong in different dimensions will miscommunicate about AI in predictable ways.

The overconfidence multiplier

The problem compounds when you factor in self-assessment accuracy. Across 184 people who predicted their own scores before taking the AISA assessment, the average predicted score was 62.2 while the average actual score was 41.9 — a gap of 20.3 points. Engineering professionals predicted 68.9 but scored 49.4 (a 19.5-point gap). Leaders who believe their teams are AI-fluent based on self-report are almost certainly overestimating.

This means the unevenness you think you have is probably different from the unevenness you actually have. You can't fix a distribution you haven't measured.

How to Actually Find Out

You have two paths: an informal self-check using the five questions above, or a structured assessment that gives you comparable data across teams.

The informal path

Run the five questions as a team exercise. Have each team lead rate their group on each dimension (1-10). Compare across teams. Where do you see the biggest variance? That's your priority.

This is cheap and fast. It's also subject to the same overconfidence bias we just described. Treat it as a directional signal, not a measurement.

The formal path

A structured team AI assessment gives you actual scores across individuals and teams, broken down by dimension. You can see where the gaps are, who's overestimating, and which teams need targeted support versus broad upskilling.

AISA's assessment uses a conversational format — candidates talk to an AI facilitator while a separate AI evaluator scores independently across 11 criteria. It detects gaming behaviours like copy-paste and suspicious speed shifts. The output is a per-person score across all five dimensions, which rolls up into team-level analytics.

Other options exist. Deloitte's 2024 Global Human Capital Trends report found that 73% of organisations planned to increase investment in workforce AI skills assessment. The market for this is growing. What matters is that you pick something that measures demonstrated capability — not self-report, not quiz scores, not tool usage logs. For a comparison of approaches, see our breakdown of 9 assessment methods.

What to do with the results

Once you have data, the playbook is straightforward:

1. Identify the floor — which teams or individuals are below the Competent threshold (score 5-6 per criterion, or roughly 60 composite)?
2. Map the variance — where is the gap between your highest and lowest teams widest?
3. Prioritise by dimension — if Safety is universally low, that's a policy and training problem. If Prompting varies wildly, that's a coaching problem.
4. Re-measure quarterly — fluency changes with practice. A single snapshot tells you where you are; repeated measurement tells you if you're moving.

The goal isn't to get every person to Expert. It's to get your organisation to a point where AI fluency differences between teams don't create invisible friction — and where everyone clears the safety bar.

---

Related reading: How Good Am I at Prompting? [Self-Test] — the individual version of this question, with a practical self-check framework.

Related reading: AI Fluency vs AI Literacy: Why the Difference Matters [2026] — why measuring what people can do beats measuring what they know.

Related reading: AI Know-How Test: Do You Know AI, or Do You Know How to Use It? — the knowledge-vs-skill distinction applied to assessment design.

Frequently Asked Questions

Is my company AI ready?

Most companies have AI access but lack AI readiness. Readiness means your teams can use AI tools effectively, evaluate outputs critically, and operate within clear safety boundaries — not just that you've purchased licences. The five-question check above, mapped to Prompting, Critical Thinking, Technical Understanding, Workflow, and Safety dimensions, gives you a fast diagnostic.

How do I assess my team's AI skills?

Start with the informal approach: ask each team lead to rate their group on the five dimensions and compare across teams. For reliable data, use a structured AI fluency assessment that measures demonstrated capability through conversation or task performance — not self-report surveys or multiple-choice quizzes. Look for tools that give you per-dimension breakdowns so you can target training where it matters.

What does AI readiness actually mean?

AI readiness means your people can identify where AI adds value in their workflows, use it to produce quality outputs efficiently, verify those outputs before they ship, and handle data responsibly. It's a capability measure, not a procurement measure. Across 1,082 AISA assessments, the average composite score is 48 out of 100 — suggesting most professionals are still in the Developing tier, regardless of how many AI tools their organisation has deployed.

Why is uneven AI fluency across teams a problem?

When one team operates at a significantly higher AI fluency level than the teams it hands work to, the receiving team becomes a bottleneck. This creates invisible drag on cycle times and frustration on both sides. Uniformly low fluency is actually easier to address because you can apply a single training strategy. Uneven fluency requires targeted, team-specific interventions based on actual measurement.