Delta

For leaders6 min read

The AI maturity model: four stages, and how to tell where your team is

A plain-English AI maturity model: four stages from Tutorial Hell to Builder, plus a quick way to place your team and pick the next step.

Newan VinthusaFounder, Delta Camp
Illustration of four blocks arranged in rising stacks from left to right, with a timer beside the shortest stack and a folder beside the tallest.

Ask a team that calls itself advanced what runs without anyone touching it, and the room usually goes quiet. That one answer places them on an AI maturity model faster than any survey does.

This post sets out four stages, then shows you how to place your own team on them before the week is out.

Most writing on the topic is aimed at enterprises with a data platform and a steering committee. I'm writing for the marketing manager or operations lead who wants to know where a team of five or fifty really sits.

What an AI maturity model actually measures

A maturity model measures how deeply AI is built into the work, not how many people have tried it. That difference matters more than it sounds.

MIT CISR built a four-stage enterprise model and found that companies in the first two stages had below-average financial performance, while those in stages 3 and 4 were well above average. I won't paraphrase their stage definitions from memory. The finding is the useful part: the gap sits between stages, not between companies that use AI and companies that don't.

Using AI is now the baseline. Stanford HAI reports that global organisational adoption rose from 55% in 2023 to 88% in 2025.

Yet McKinsey's global survey found only 37% of organisations say AI has contributed to their EBIT. Almost everyone is using it. Far fewer are being paid back for it.

The four stages, as I score them

These are the stages in Delta Camp's AI Maturity Audit, individual track. Four scored questions give a score from 0 to 12, and the score maps to a stage. There is no AI model doing the scoring, just arithmetic.

  • Tutorial Hell (0 to 3): You know more about AI than your work shows. You've watched hours of videos and read the threads, but no task in your week has changed.
  • The Dabbler (4 to 6): You've had wins, and none of them stuck. Nothing runs unless you run it yourself.
  • The Operator (7 to 9): Things run without you. The next job is making that count for the whole team.
  • The Builder (10 to 12): You're past learning and you should be shipping. You are capable, but you haven't yet finished a thing with customers attached.

The expensive part of the first stage is the distance between knowing and shipping. Nobody sits there for lack of intelligence. They sit there because nobody made them pick one task and finish it.

How to tell which stage your team is in

A team is never one stage. It's a spread, and the average hides the useful bit.

Put these four questions to each person in one team meeting:

  1. Name one task that ran this week without you starting it. A blank look means Tutorial Hell or Dabbler.
  2. Say who owns that automation and who would notice if it broke. No owner means it's a Dabbler win, not an Operator one.
  3. Tell me where the saved time went. If nobody can say, you have the problem I describe in the capture gap.
  4. Explain who checks the output and how long that takes. Slow checking quietly eats the saving, as I cover in the four-hour tax.

Tally the answers and you'll see your spread. If you find an Operator or two, they are the first people to ask for help.

Why the business version scores six areas

The business track of the audit takes about 6 minutes. It scores six areas, from the foundations up: data, processes, people, AI in practice, governance and risk, and strategy and value. You get one of three readings: Foundations first, Ready for targeted pilots, or Ready to scale.

It works this way because a lot of 'AI isn't sticking' turns out to be messy data or undocumented processes, not an AI problem. In BCG's survey of employees across APAC, only 57% said their organisation is redesigning workflows to fit AI, compared with 70% of global businesses at the forefront of adoption. If your processes live in people's heads, a new tool has nothing to attach to.

People matter just as much. Gartner's survey of organisations in the U.S., U.K., France, Germany, India and Japan found that business units trust and are ready to use new AI solutions in 57% of high-maturity organisations, against 14% of low-maturity ones. In my experience that trust follows watching one small automation work on real work.

At the top end, Deloitte's global survey of senior leaders found only 25% use AI to deeply transform the business through new products, reinvented core processes or new business models. The last stage is rare, so don't feel behind if you haven't reached it.

What to do at each stage

  • In Tutorial Hell, stop watching and audit your own week: list recurring tasks, put a time on each and pick three. Audit before tools walks through the method.
  • As a Dabbler, take the one win that worked and make it run on a trigger, so it no longer depends on your afternoon.
  • As an Operator, write down who owns each automation and how its output gets checked, then decide where the freed time goes.
  • As a Builder, finish one thing that a customer or colleague depends on, instead of starting a fifth.

This week, take the free AI Maturity Audit yourself (about 3 minutes for the individual track), then ask one colleague to take it too. Compare your stages over coffee, because the gap between your two answers is your team's real starting point.

Frequently asked questions

What is an AI maturity model?

It's a way of placing a person, team or company on a scale that shows how deeply AI is built into the actual work. The useful ones measure what runs and who owns it, not how many people have tried a tool.

How many stages does an AI maturity model have?

It depends on who built it. MIT CISR's enterprise model has four stages, and Delta Camp's individual audit also uses four. The business version gives three readings instead, because a company's problems often sit in its data and processes.

How do I know which AI maturity stage my team is in?

Ask each person to name one task that ran this week without them starting it, and who owns it. Their answers show you the spread across the team. The free AI Maturity Audit does the same job in a few minutes.

Is it bad to be at an early AI maturity stage?

No. Early stages are where most people start, and the fix is usually one real task finished rather than more learning. The risk is staying there while telling yourself you're advanced.

How long does the AI Maturity Audit take?

The individual track takes about 3 minutes and the business track about 6. It's free, there's no call, and your result shows on screen and arrives by email.

Where the numbers come from

Keep reading

Illustration of a stack of notes beside an empty tray with a pen laid across it and a timer standing between them.

For leaders6 min read

AI literacy training for employees: what changes how people work, and what wastes the budget

What actually changes how people work after AI literacy training for employees, and where the budget gets wasted, with APAC data.

Read the post
Illustration of an hourglass leaking orange sand beside four blocks of equal height: a chart that isn't growing.

For leaders6 min read

The capture gap: why your team saves hours with AI and your company doesn't

AI saves people real time, yet most companies see no change in their numbers. Where the savings leak, and the workflow decision that captures them.

Read the post