The Real Story Behind Your AI Adoption Number

Key Takeaways

• Adoption Isn't Impact: Adoption tells you AI is present, not what it's changing.
• Four Layers, Not One: AI use splits into day-to-day, management, strategy, and ecosystem layers.
• Most Activity Is Operational: PeopleCert's David Cannon estimates the vast majority of current AI use remains concentrated in Layers 1 and 2.
• Cloud Already Ran This Experiment: Only 10% of cloud transformations ever hit their full projected value.
• ITIL Adds Structure: The AI Capability Model gives teams a shared language for what AI is doing, then connects that to value and risk.

A department head stands up in a leadership meeting and reports the good news: adoption is universal across the company. The dashboards are green. Every team has tools in hand and is using them daily.

Then someone in the back asks a simple follow-up: so what's changed about how we compete? The room goes quiet, because nobody knows.

Adoption measures how many people are touching AI, not what that AI use is doing for the business. McKinsey's 2025 State of AI survey found that 88% of organizations now use AI in at least one business function, a figure that sounds like transformation until you ask what layer of the business that use is changing.

AI use isn't one thing. It's at least four different things happening at once, and organizations are often running almost entirely on the day-to-day layer.

Cloud computing followed this same curve, and last time, the mismatch between adoption and strategy cost companies years. The four layers show where that mismatch appears, and why reaching strategic or ecosystem-level use takes a deliberate choice, not just more adoption.

One Number, Four Different Stories

AI use inside most organizations isn't one activity. It splits into four distinct layers, and each one changes something different about how the business operates.

Executives tend to talk about AI use as a single number, an increasing percentage reported alongside other adoption metrics every quarter. But that number collapses four different kinds of work into one figure, and conflating them is exactly how a leadership team ends up mistaking movement for strategy.

Here's what separates them:

• Day-to-day execution: individuals and teams using AI to get their own work done faster
• Management optimization: AI applied to coordination processes like ticket routing and resource allocation
• Strategy: AI used to build new products, models, or markets
• Ecosystem change: AI reshaping relationships across customers, partners, and suppliers

Adoption tells you how many people across those four layers are using AI at all. It says nothing about which layer they're using it in, and the distinction matters. A company can report near-universal AI use and still be doing almost all of it at the first layer, with individuals working faster while the business itself stays the same. Layers one and two are worth a closer look first, since that's where the majority of AI use in most organizations falls.

Faster Work Is Still the Same Business

Two layers absorb almost everything that gets called AI use today, and both make things run faster without changing what's actually happening.

Executives sometimes assume that heavy AI use signals a shift in how the business operates. More often, it reflects teams and processes running the same playbook with less friction. The difference between the two layers comes down to scope, not sophistication.

Here's how that scope plays out:

• Layer 1 (day-to-day execution): lives inside one person's task list, showing up as time saved rather than outcomes changed.
• Layer 2 (management optimization): shifts the frame to how work gets distributed and tracked across a team, showing up as a faster queue or an earlier flagged escalation.

The scope is wider at Layer 2, but the goal matches Layer 1's, running the existing workflow better. Most current AI activity lives here: Cannon puts Layer 1 alone at 80% to 85%, with another 10% to 15% at Layer 2.

Combined, that accounts for nearly all AI activity most organizations report today, and both layers can deliver real gains inside the business that already exists. The next two layers go beyond that, into building something new.

Where Strategy Actually Starts

Past the two layers most organizations live in, AI starts doing something different. It stops making the current business run smoother and starts changing what the business is, or what it can offer that didn't exist before.

The shift beyond optimization plays out in two distinct ways, differing in how wide the change actually reaches.

Here's what separates the two:

• Layer 3 (strategy): changes what one organization offers or does, building products, business models, or markets that didn't exist for it before.
• Layer 4 (ecosystem change): changes how multiple participants in a market create or exchange value, reshaping relationships between customers, partners, and suppliers.

Cannon puts Layer 3 (strategy) activity at 5% of AI use or less, industry-wide, and notes that most of it belongs to startups built specifically to deliver AI tools. Everyone else is still absorbed in day-to-day operations and management.

Layer 4 is rarer still. Outside of AI-native startups, clear examples at this layer barely exist yet, and that absence is part of what the framework is pointing at. A category can exist before the market has filled it in.

Between them, Layers 3 and 4 account for a small fraction of what most organizations call AI use today. Cloud computing followed a similar pattern, individual use first, then broader organizational adoption, then a handful of companies that used it to compete on an entirely different scale, and what happened to the rest of the field along the way is worth a closer look.

What Cloud Infrastructure Taught Us

AI isn't the first technology whose widespread adoption has outpaced its strategic use. Cloud computing followed a similar pattern roughly a decade earlier, and the mismatch it left behind is still visible in the numbers today.

Adoption outran strategy then too, and the consequences showed up less as failed projects than as a steady drain. McKinsey's research on enterprise cloud value found that only 10% of cloud transformations achieve their full projected value, a decade into a technology most of those organizations had already adopted.

Most organizations adopted the technology without ever closing that gap. Cloud adoption had to clear a baseline of infrastructure and procurement to begin, friction points AI can sometimes bypass.

Cannon's assessment is that AI's version of the same problem could be ten times worse or more, since AI tools reach employees directly, often without the IT oversight cloud tools required from the start.

AI is tracing the same early shape. What determines whether an organization repeats this pattern or breaks it has less to do with how much AI it adopts than with what kind of questions it asks of the technology.

More AI Isn't More Strategy

More AI adoption doesn't automatically produce more strategic AI use, and that's easy to miss when both get reported under the same headline number. The two aren't stages of the same journey. They're different bets, and increasing the size of one doesn't change the size of the other.

Adding more users and more tools can increase Layer 1 activity without creating Layer 3 or Layer 4 activity, because strategic AI use isn't a function of volume. It's a function of the questions someone asks. Each layer starts from a different one.

Here's how that question changes at each layer:

• Layer 1: asks how this task can be done faster.
• Layer 2: asks how this process can run better.
• Layer 3: asks what the organization could now offer or do differently.
• Layer 4: asks what new relationships or markets this makes possible.

Most organizations have plenty of people asking the first question and few asking the third or fourth.

Layer 1 and Layer 2 work isn't lesser for any of this. The goal isn't to replace operational AI use with strategic AI use, since an organization can pursue value at multiple layers at the same time. Automating a mundane task is often exactly the right use of AI. The risk isn't using AI there, it's mistaking a pile of those wins for a strategy the organization never built. With only 6% of our audience having a fully embedded and documented AI strategy, getting there consistently takes a framework, not good intentions alone.

Where Ambition Meets Structure

Building an organization that consistently asks Layer 3 and Layer 4 questions takes more than knowing they exist. ITIL provides the structure to create consistency, giving organizations something more concrete than a vague mandate to "be more strategic."

The four layers answer one question: what is AI use changing. ITIL's AI Capability Model, known as the 6C model, answers a different one: what is the AI doing within a given use case.

It breaks AI use into six capabilities, creation, curation, clarification, cognition, communication, and coordination, giving teams a shared vocabulary for describing what it's doing rather than which product it came from. The same capability can show up at any layer. Creation might mean an employee drafting an email at Layer 1, or it might power a new AI-enabled product at Layer 3.

The shared vocabulary is the starting point. Once a team can name what an AI capability is doing, two things follow.

• A line back to strategy: the named capability can now be tied to a valuable outcome the organization is trying to achieve, not just a task it automates.
• A way to surface risk: identifying the capabilities involved gives teams a more precise way to examine the risks tied to that outcome.

Deciding to pursue a Layer 3 or 4 opportunity is still the organization's call to make. What ITIL changes is everything that happens next, turning ambition into something a team can actually build.

Know What Your AI Use Is Building

Two companies, both reporting heavy AI use, both showing similar adoption numbers on paper. One is building new revenue lines. The other is only running its existing operations a little faster.

Adoption tells you whether people are using AI. It doesn't tell you what that use is changing. A company can be doing almost everything right at the day-to-day and management layers while nobody asks what AI could let the organization become.

Looking at AI adoption alone shows an incomplete signal. It tells you AI is present, but not which of the four layers that presence is reaching.

New Horizons' ITIL training gives teams the shared language and structure to recognize, evaluate, and apply AI capabilities against business opportunities tied to strategic outcomes.

Where is your organization's AI activity concentrated, and is anyone deliberately asking the Layer 3 or 4 questions yet?

Explore New Horizons' ITIL training and find out what your organization's AI use is building.