A benchmark of more than one hundred Lean, Lean Six Sigma and Quality Improvement practitioners suggests that artificial intelligence has arrived on the desk more readily than it has reached the work. The tools appear to be within reach in most of the organizations these practitioners describe, and few report any shortage of interest in them. The use they describe, however, tends to be a narrow one, in which AI drafts correspondence and captures meeting notes far more often than it maps a value stream or works through a root cause. On the evidence of this sample, the practitioners whose discipline is to improve how work is done are, for the most part, not yet applying AI to that work. The pattern points less to a problem of access or appetite than to one of practice, the still-developing skill of directing AI at the work that matters and remaining accountable for what it returns.
Adoption is wide but shallow
The benchmark's first observation is that access no longer appears to be the binding constraint. Among practitioners who reached the AI questions, roughly half describe AI in their day-to-day operations as piloted or confined to a few specific areas rather than widespread, and a further group describes early, exploratory use. About one in five describe operations that remain largely manual, with AI neither in use nor being explored, while the share reporting AI in regular operational use across multiple parts of the organization stands at six percent. The distribution is wide but shallow: most respondents have moved into some form of use, and comparatively few have carried it through to the point where it is embedded in how the organization runs.
This pattern is worth pausing on, because it narrows the range of plausible explanations. Where a technology clusters at the unstarted end of a scale, the limiting factor is usually access, cost, or permission. The benchmark suggests a different situation, in which the tools are present, pilots are underway, and interest is evident. What the data leaves open is not whether practitioners can reach AI, but what the work gains once they have.
What AI is mostly used for
The responses describing what AI is used for cluster, with few exceptions, around general office work. The uses mentioned most often are capturing meeting notes and minutes, drafting and editing written material, summarising documents, preparing reports and briefings, and refining language, with Microsoft Copilot named more frequently than any other tool. These are legitimate and useful applications, and they help to explain the speed of adoption, since they ask little of the surrounding system and return an immediate, visible saving of time.
The more telling pattern is what the responses rarely mention. Of about one hundred respondents who wrote about how they use AI, two referred to the improvement work that defines their discipline, the value-stream mapping, root-cause analysis, standard work, and structured problem-solving that Lean, Lean Six Sigma and Quality Improvement practitioners are trained to lead. The sample is, on its own terms, a practitioner sample, since every respondent holds a belt. On this evidence, the instrument that might be expected to accelerate the craft is instead being applied to the correspondence that surrounds it. AI appears to have been adopted as an assistant to the work around the work, rather than as an instrument of the improvement itself.
The work it has not yet reached
This gap does not appear to be a failure of imagination. Asked to name an operational problem that AI might help with but had not yet been applied to, respondents answered readily and in concrete terms, and their answers point toward the operating core that current use tends to avoid. They described manual data entry, validation, and the movement of information between disconnected systems; scheduling, staffing, and rostering; the routing and triage of intake and referrals; and the tracking of compliance across multiple sites. These are not peripheral conveniences but the recurring, load-bearing problems of running an operation, and they are precisely the problems a trained improver is equipped to frame.
The reading that follows is not one of practitioners who cannot see the opportunity, since they describe it with some precision, but of practitioners who have not yet connected the tool already on the desk to the work they are trained to diagnose. AI sits on one side of that distance, applied to writing and notes, while the improvement work sits on the other, named in detail as the place it might go.
A smaller, related signal sits beneath these answers. Where respondents explained what was holding wider use back, the theme raised most often was not cost or technology but direction, in that tools were available, several observed, with little guidance on when or how to use them, and individuals were left to work it out for themselves. Running alongside this was a genuine and appropriate caution about confidentiality and the responsible handling of sensitive information. Taken together, these read less as reasons to deploy more quickly than as indications that the missing element is capability, the ability to apply AI to a given task and to judge and stand behind what it produces.
A question of practice, not tools
A consistent theme in LEA's work is that technology tends to amplify the system it enters. Applied to a disciplined process, AI can compound the discipline; applied to a process that is unclear or unstable, it tends to produce faster and more confident versions of the same problems. The same principle holds one level down, at the level of the practitioner. In the hands of someone able to frame a problem, separate a cause from a symptom, and remain accountable for a conclusion, AI can act as an amplifier of judgement, whereas in the hands of someone reaching for it without that frame, it becomes a quicker route to plausible work of uncertain quality.
For that reason, the movement from assistant to amplifier reads as a question of skill rather than of software. It asks for practitioners who can select an appropriate form of AI for a given improvement task, direct it with a well-formed prompt, incorporate it into the cycle of structured problem-solving without ceding the judgement that cycle depends upon, and examine what it returns for the errors and gaps that confident output can conceal. Little of this is the tool's to provide. It belongs to the practitioner, and it can be taught.
What appears scarce is not the tool, but the practice of aiming it at the work that matters.
An opening, not a shortfall
The benchmark describes a field that has acquired AI more quickly than it has learned to apply it. For the practitioners who lead improvement, that reading is closer to an opening than a shortfall, since the tools are already within reach, the opportunities are already visible, and the element still to be developed sits within their existing discipline. Building that capability is the purpose of LEA's AI-Ready Lean Practitioner program, a focused enablement intended to help Lean, Lean Six Sigma and Quality Improvement practitioners apply AI to the improvement work itself, with judgement and accountability intact.
As with every reading from this benchmark, the figures describe the field rather than any single organization within it. Knowing where a particular operation sits, not on AI alone but across the operating system that determines what AI can amplify, is the purpose of an OpsScan. The benchmark indicates what appears to be true of the sector; an individual assessment indicates what is true of a given organization.
Adoption figures are drawn from the LEA OpsScan benchmark, a self-assessed survey completed by more than one hundred Lean, Lean Six Sigma and Quality Improvement practitioners who finished the full assessment, with AI adoption rated on a one-to-five scale. Open-text figures are qualitative themes drawn from those who answered the AI questions, and are not presented as statistics. Respondents are individual and anonymous; the sample is self-selected among improvement-engaged practitioners, Ontario-weighted, and the findings are directional and strengthen as the benchmark grows.