The AI maturity pyramid: climb, don't leap
Organisations that stall with AI often try to jump straight to the top. The AI maturity pyramid describes four levels, each built on the one below. Every AI strategy I deliver starts by establishing which level you are on today and what you need to build to reach the next one.
The framework is by Vivianna Fang He (UCL) and Phanish Puranam (INSEAD). Source at the end of this page.
Four levels, one foundation
Select a level to see what it looks like, where it stops paying off and how to reach it.
Level 1 of 4
Individual productivity
People use AI to draft, summarise, analyse and code faster, with less effort.
Examples. Coding assistants, first drafts of documents and campaigns, desk research.
The limit. Gains stay with individuals, and competitors can buy the same tools. The result is parity, not advantage.
How to get there. Give people approved tools and clear guidance, build AI fluency across roles, and make experimenting a normal part of work.
Where I help. AI training to build fluency across roles, and AI strategy for the adoption plan and guardrails.
Level 2 of 4
Group productivity
AI moves into the way teams work together: meetings, handovers, routing work between departments, and answers drawn from the organisation’s own knowledge.
Examples. Meeting summaries with tracked actions, handover templates, an internal knowledge assistant.
The limit. Still easy to copy. The organisation works better, but the business itself has not changed.
How to get there. Turn individual wins into dependable team routines and automate the repeatable tasks around them.
Where I help. AI training to turn individual habits into team routines, and process automation build for the first workflows.
Level 3 of 4
Business process efficiency
AI runs inside core operations, cutting cost and errors, speeding up response and making scale possible.
Examples. Automated underwriting, demand forecasting, predictive maintenance, invoice processing.
The limit. Mostly defensive. You compete better within your industry as it is, but customers value the same things and revenue comes the same way.
How to get there. Redesign core processes on the fluency and routines built at the first two levels, with data you can trust and a named owner for every process.
Where I help. Data foundation to make the data ready, and process automation build to put AI into the process.
Level 4 of 4
Business model transformation
AI changes how you create and capture value, not only how the work gets done. Advantage rests on data, redesigned workflows and people, which are hard to copy; algorithms increasingly are not.
Examples. Rolls-Royce sells engine uptime, paid by the hour, rather than engines, and AI-enabled monitoring makes that viable at scale. DBS Bank built data and AI capabilities for years before launching marketplaces for cars, property and travel.
The risk. It cannot be mandated. When ambition runs ahead of capability, strategy loses touch with execution.
How to get there. A few selective, high-impact bets, made once the levels below are in place.
Where I help. AI strategy to frame and test the options, and part-time AI leadership to see them through.
Every level rests on the same foundations: data you can trust, model governance, clear decision rights, a named owner for each process, and metrics that tell real impact from novelty. Weak foundations are the most common reason AI stops at a pilot, which is why a data foundation review is often my first piece of work.
Why organisations stall
Leaping to the top
A business model built on AI is the goal many leaders announce first. Without broad AI fluency and AI already working inside operations, that ambition loses touch with execution, and without people's buy-in the change does not stick. Level 4 is reached by mastering levels 1 to 3, not by skipping them.
Choosing between top-down and bottom-up
It is a false choice. Leadership sets the conditions: literacy, infrastructure, guardrails and clear success criteria. Within them, good ideas come up from the people doing the work. A strategy needs both, and someone who owns the climb to the next level.
Where are you today?
Tick each statement that is true for your organisation as a whole, not just one team. It takes about two minutes.
Your starting point
Before level 1
Build next. Approved tools, clear guidance on data, and time for people to learn.
Where I help. AI training to build the basics, and AI strategy for the adoption plan and guardrails.
You have ticked statements above your current level. That is common, and it is where initiatives tend to stall: close the gap below before scaling what sits above it.
This self-check is my own simplified reading of the pyramid, not an official or validated instrument. A level counts as reached when at least two of its three statements are true and every level below it is reached too. Your answers stay in your browser; nothing is sent or stored.
How I assess maturity in an engagement
The assessment is the first step of an AI strategy, in the Frame stage of how I work.
01
Listen
Interviews with leaders and with the people doing the work, to see how AI is used day to day.
02
Measure
A short staff survey, plus a review of current AI use cases, data, governance and ownership.
03
Plan the climb
Your current level, the gaps holding you there, and a roadmap to the next level with owners and metrics.
Source: Vivianna Fang He and Phanish Puranam, “The AI Maturity Pyramid”, INSEAD Knowledge, June 2026. Company examples are from the article.
Not sure which level you are on?
Book a short call. I will give you a straight read on where you are and what to build next.
Book a call