An AI maturity model describes the stages a company moves through as AI goes from novelty to operating capability. Consultancies publish elaborate versions with spider charts and forty dimensions. Useful for enterprises; mostly noise for a growing business. What an operator needs is a model honest enough to place yourself in and specific enough to tell you what to do next.

Here's the five-level model I use with clients — the same dimensions scored by my free AI readiness assessment (data, process, people, governance), collapsed into stages.

Level 0 — Unaware or avoiding

AI isn't on the leadership agenda, or it's been dismissed as hype. Meanwhile individual employees are almost certainly using it anyway — which means you're not actually at level zero; you're at level one without knowing it.

The tell: "We haven't really gotten into AI yet." The move up: Leadership fluency, not tools. A briefing, a competitive scan of what your industry is automating, and an honest look at where hours go in your own operation.

Level 1 — Ad hoc experimentation

Individuals use AI tools on their own initiative — drafting, research, code — with no coordination, no policy, and no shared learning. This is where most small and mid-sized companies actually are, and it's also peak shadow AI exposure: all of the data risk, little of the compounding benefit.

The tell: Enthusiasm in pockets; nobody can list what tools are in use. The move up: An AI audit to surface reality, a one-page policy to make safe use legal, and a named owner for AI decisions.

Level 2 — Sanctioned adoption

Approved tools on company accounts, a written policy, and an owner. Teams share what works. AI is saving real hours, but the wins are individual productivity, not process change — faster emails, not fewer emails.

The tell: "We use AI" means everyone uses a chatbot well. The move up: Pick one measured process automation — one workflow, an ROI target, a 90-day window. This is the jump from using AI to operating with it, and it's where most companies stall or fail by skipping the measurement.

Level 3 — Process integration

AI runs inside actual workflows — lead handling, quoting, support triage, reporting — with human review at defined points and ROI measured per initiative. A roadmap exists; things that miss their targets get killed.

The tell: You can name the processes AI runs, what each saves, and who owns each one. The move up: Portfolio discipline. Sequenced initiatives, quarterly governance review, and executive-level ownership — this is typically where dedicated AI leadership (full-time at enterprise scale, fractional below it) earns its keep.

Level 4 — Compounding capability

AI is how the business operates, not a program within it. New automation builds on existing rails; governance keeps pace as usage scales; the company evaluates emerging capabilities (agents, deeper integrations) from a position of strength rather than fear of missing out.

The tell: Automation proposals cite existing infrastructure. The board asks about AI the way it asks about finance — routinely, with real answers. The maintenance work: Keep the kill discipline. Mature companies accumulate zombie automations exactly the way immature ones accumulate unused SaaS.

How to use this honestly

Two rules. Place yourself by your weakest dimension, not your proudest anecdote — a company with brilliant ad hoc use and zero governance is Level 1, whatever the demo looks like. And don't skip levels: jumping from Level 1 to a Level 3 automation project on ungoverned data and undocumented process is the standard recipe for a six-figure failure.

Three minutes with the readiness assessment gives you a scored placement and your weakest dimension. If the result puts real money at stake, a Strategy Intensive turns the placement into a sequenced plan for the next two levels.