Originally published in Fortune
ActivTrak’s Productivity Lab tracked 120,620 employees over three quarters and found something counterintuitive: the optimal level of AI adoption maturity for most employees may be somewhere in the middle between shallow AI usage and full automation.
Most leaders I know are tempted to build their AI adoption strategy as if every employee should be an AI super user. Buy the most powerful tools, push everyone toward the deepest integration, maximize adoption maturity and assume productivity will skyrocket.
The most recent data from ActivTrak’s Productivity Lab complicates that idea. Productivity and work-health metrics rise as employees move from little or no AI use to regular, task-level adoption, with healthy utilization peaking at 75%. But once AI becomes embedded in workflows, healthy utilization drops about 5 percentage points — to levels statistically indistinguishable from employees who barely use AI.
The right level of AI adoption maturity depends on the work being done and the business objectives it supports.
Why moderate AI maturity may be sufficient for most
Traditional AI adoption metrics track licenses or login counts, which measure deployment but offer little clarity into how AI impacts the work being done. Typical AI maturity models measure an organization’s overall progress toward deeper AI adoption. Our Productivity Lab takes a different approach, using behavioral data to document how AI actually changed the way people work, and categorizing them into three stages of maturity that reflect true operational progression.
The Lab tracked the same 120,620 employees across 1,009 organizations for three consecutive quarters from Q4 2025 to Q2 2026. The data showed 27% of employees used AI like a search engine to answer questions and summarize information (Stage 1, Research Assistance). 14% used AI to draft content, generate ideas and complete routine tasks that they then validate and finalize (Stage 2, Task Execution). Only 2% reached the stage where AI becomes an integral part of day-to-day workflows (Stage 3, Workflow Integration). Overall, AI users remain a minority at 43% of employees studied.
Stage 2 is where AI eliminates repetitive work — for example, allowing a sales rep to generate a quote from five systems with a single prompt instead of manually compiling the information. That’s where healthy utilization peaks, and where most organizations should be focused.
AI consumption is not equal to AI maturity
Most AI maturity models were built to reward consumption. They aim to qualitatively measure how much, not how well. As a consequence, they tend to reinforce the perception that “most” usage is best. That assumption leads to two specific risks.
One, organizations may lose sight of runaway costs. More mature usage means more powerful models, more tokens, more infrastructure. If the task or role doesn’t require it, you’re just spending money you could invest elsewhere.
Two, organizations may unwittingly foster operational disconnect. Employees sprinting ahead may generate sophisticated AI workflows that optimize individual tasks but without improving broader processes. If the workflow hasn’t been redesigned around business goals, all you’ve done is produce more AI slop, faster.
In both cases, the answer is more visibility into how AI transforms work for your organization.
Why most organizations are stuck in high adoption and low maturity
- Organizations roll out AI tools without investing equally in guidance for employees.
Here’s an example from ActivTrak: When our operations team noticed Anthropic costs rising, they dug into the data to understand why. They discovered employees routinely using the newest, most powerful model to rewrite customer emails — a task that didn’t require that level of sophistication. That led us to create an internal reference to help employees match the right model to the right task. The problem wasn’t the output; it was defaulting to the most powerful model for every job — the AI equivalent of hiring a superstar to do work that didn’t need one.
- Organizations sprint to pilot AI without first slowing down to see how work currently flows.
Before implementing AI tools, organizations must take the time to map workflows — so the right investments are being made in the right places. Consider how work happens now, and what would change if AI did the work. Traditional business principles still apply: everyone fits a role and everyone has strengths and weaknesses.
Resist the urge to implement one massive department-wide initiative in favor of addressing small workflow fixes. Build, measure and learn to find the optimal sweet spot.
The durability finding — why the target decision is consequential
Defining maturity targets is a leadership responsibility because AI adoption is a durable change. Productivity Lab data shows 82% of employees who adopted AI kept using it. But once people move past casual use, they continue to use it quarter after quarter, and almost no one who goes deep ever comes back.
The level of adoption I push my team toward is the level of adoption they’ll likely stick with. So if I drive everyone to the deepest tier, I may be locking them into usage where productivity gains stall out and costs exceed benefits.
The question isn’t only how we increase the 2% who integrate AI into workflows. It’s also how to coach the 27% of novice users to reach task assistance fluency.
The prescription — right tool, right role, right stage
We need to cut through the noise that’s in the market right now to get to a more level-headed place about what problems we’re trying to solve. No single maturity model or AI tool should define AI strategy. Strategy depends on the makeup of your company, the type of work you’re doing and the goals you’re trying to achieve. A lean AI-native company may need most people functioning alongside embedded AI workflows and agents, whereas an established business may find AI task assistance is more than enough competitive advantage and requires less operational disruption.
Discipline is critical. Without intentionality, maturity often brings unmanaged costs, misaligned workflows and locked-in behaviors that are rarely reversible. The model companies want you to consume is as much AI as possible. Our job as leaders is to not dismiss the fact that AI can be a competitive advantage, because it absolutely is. But it’s only a competitive advantage if you know how to accurately invest in and use the right tools to do the right job.
