The framework

The AI Maturity Matrix.

Five levels of AI maturity, read across six dimensions, grounded in our primary research.

The five levels we observe in our research.

Each level builds on the one before. The definitions are drawn from our ongoing research into how companies adopt AI in practice.

Level 0

Wild West

Where everyone starts: energy everywhere, ownership still forming.

SignalsA curious few experiment, and clever work appears in pockets.

ToolsA scattered mix of company and personal tools, chosen person by person.

AutonomyChat, mostly. Occasional sparks, no automation yet.

Next step upAn owner: one person who turns sparks into progress.

Level 1

Prioritized

AI has an owner and real goals.

SignalsEarly adopters emerge, and the first power users appear.

ToolsA sanctioned tool list forms; prompts travel person to person.

AutonomyMostly prompting at solo scale, with the first repeatable workflows.

Next step upA funded plan behind the goals.

Level 2

Adopted

A funded plan is working: most people use AI most days.

SignalsPower users pull ahead, and shared skills start to accumulate.

ToolsAn official skill list appears, often a spreadsheet, and grows fast.

AutonomySkills and the first sub-agents; power users pick their own models.

Next step upA dedicated team to run AI as an operation.

Level 3

Operationalized

A dedicated AI ops team runs skills as products.

SignalsPower users multiply, and governance has real teeth.

ToolsCentral tools spread the best practice to every team.

AutonomyAgents run real workflows, with central review behind them.

Next step upWiring every function.

Level 4

Instrumented

AI is wired into the org itself. AI engineers sit inside functions.

SignalsEveryone is a power user; the work can't be done without AI.

ToolsBest practice is built into the rails and improved by the system itself.

AutonomyRoutines run on their own; people direct, review, and steer.

At the topThe system keeps learning.

Place yourself on the matrix.

Four diagnostic questions, all about how you manage ownership. Answer them in order about your own company: every yes is a step up.

Question 1

Owner

Does someone own AI adoption today? A yes is Level 1 or above.

Question 2

Plan

Is there a funded plan behind the goals? A yes reaches Level 2.

Question 3

Team

Is a dedicated AI ops team in place? A yes reaches Level 3.

Question 4

Wiring

Has that team wired the whole org? A yes is Level 4.

The full matrix: six dimensions, ownership first.

In our research, ownership is what enables the rest: the other five dimensions follow it, and they are the same ones the AI-Use Score measures.

Dimension	L0 · Wild West	L1 · Prioritized	L2 · Adopted	L3 · Operationalized	L4 · Instrumented
OwnershipDefines the level	Personal energy, personal tools	An owner, with real goals	An owner and a funded plan	A dedicated AI ops team	The team has wired the org
Adoptionusers → power users → AI-native	A curious few	Early adopters, first power users	Most people, most days	Power users multiply	Everyone's a power user: the work can't be done without AI
Sophisticationprompts → tools → agents	Occasional sparks	Mostly basic, at solo scale	Power users pull ahead: skills, sub-agents, model choice	Central tools spread power-user practice	Best practice built into the rails
Efficiencywho manages cost	Too small to matter	The bill sometimes surprises finance	Power users self-manage cost	Cost dashboards from the central team	Managed automatically: routing, model choice
Reusewhere shared work lives	Individual know-how	Personal libraries, shared person to person	An official skill list, often a spreadsheet	Skills as products, centrally owned	Offered automatically, self-improving
Governancehow policy is enforced	Habits, not yet rules	A tool list, not a policy	A written policy, self-enforced	Central review, after the fact	Preventive: caught before it happens

Real companies are uneven. The cells describe what we most often observe at each level, not a template your business must match.

Why we drew the matrix

Mike Redbord

Co-Founder · August 2026

A company we talked to this summer runs 140 internal agents. Nobody there can tell you which one helps with what. That conversation is the reason this page exists.

We started asking every company the same questions about AI adoption: who owns it, what's the plan, who runs it, how far has it spread. We expected the answers to sort companies by sophistication. The companies with the most AI-pilled employees at the top, everyone else below. That's not what we found.

What lets a company improve, on AI adoption, sophistication, any of it, turns out to be less about how skilled individual employees are with AI and more about how the company manages AI ownership. A company with a mediocre AI toolset and a real owner keeps getting better. A company with brilliant individual work and no owner stays exactly where it is, and the brilliant work retires with whoever did it. Once we saw that, the rest of the matrix fell out of the research on its own.

Some other things we heard often enough to write down:

Top performers are the top AI users, at every company we've asked. Nobody has shown us the reverse yet.
Skill lists go stale everywhere. At one company the prompt is on v5 and the team is still running v1. Nine people were solving the same problem, each slightly differently.
Lunch-and-learns don't stick. The people who present are the people who already use AI.
One rep ran a homegrown agent for two months before anyone found out about it. Self-reporting is most companies' entire detection system.
People fork the shared skill to make it their own and never come back. The official version keeps aging while private copies multiply.
Cost data will show you who uses the most tokens. A power user should be more than a tokenmaxxer, but most companies have no data that can tell the two apart.

One more pattern: engineering teams are way ahead of GTM teams on internal AI, at basically every company we've met. We work on the GTM side anyway. Those are our people.

The levels describe what we most often see together. Plenty of companies are Level 3 on ownership and Level 1 on governance (that combination is common enough that we stopped being surprised). Read your company row by row and expect the rows to disagree.

We keep revising this as the research accumulates. If your company reads differently than the matrix predicts, I'd love to hear about it at [email protected]: that's how the levels got their shape in the first place.

Mike

Measure your level with your real AI usage data.

The 90-day pilot places you dimension by dimension from your team's actual AI usage data. See it on your own data, or talk it through with us.

Book a demo