Lean and AI in 2026: what the lean community is saying

AI is the theme of the Lean Global Connection 2026, “Designing human-centered organizations in the age of AI”, an online event of the Lean Global Network taking place on November 12 and 13, 2026: about thirty sessions carry AI in their title, from Bosch’s “Digital Gemba” to “Jidoka in the Age of AI” and Michael Ballé’s “What AI Will Never Do”. AI and technology were also one of the three themes of the Lean Summit 2026 in Houston in March, alongside leadership and management systems, where the Lean Enterprise Institute presented its new Lean Tech initiative.

The community is not yet fluent in the subject. In a July 2026 discussion published by the Lean Enterprise Institute, Art Smalley guessed that “up to 80% of people in the lean community” still have little to no understanding of AI.

The most repeated message of 2026 is that an unstable process should be stabilized before it is automated, or AI will speed up its waste. It runs from Lean Summit keynotes to Lean Global Connection sessions and to Tom Peperkamp’s February 2026 article for the Lean Enterprise Institute on optimizing the flow rather than a single step. Our article on lean before AI gathers the evidence and turns it into four questions to ask before automating a task.

The Lean Enterprise Institute has mapped prompting onto PDCA. In “Prompting for Problem Solving” (July 22, 2026), Tyson Heaton and Art Smalley write that the prompt replaces the Plan, the model does the Do, and “Check and Act stay mainly with you.” They advise naming the problem-solving method, prompting one step at a time, choosing the type of analysis yourself, and writing guardrails into every prompt; our article on AI and problem-solving skills lists those guardrails.

The harness around a model matters more than the model itself. In a February 2026 article, Art Smalley reports that when he started building lean AI tools, raw model output was “maybe 60% useful”, and that adding structured prompts, Toyota-based problem-solving frameworks and feedback mechanisms raised it to 80–90% with the same model.

As AI agents start to carry out work, lean practice reaches for jidoka: detect the abnormality, stop, make the problem visible. Toyota itself rooted its Global AI Accelerator in jidoka, according to Art Smalley (February 2026), and the Lean Global Connection has a session on “Jidoka in the Age of AI”. Our article on jidoka for AI agents explains what counts as an abnormality for an agent and how to write the stop into prompts.

Whether AI builds better problem solvers or weaker thinkers is an open question in 2026, asked word for word by a Lean Global Connection session. The answer emerging from practitioners depends on the setup: an assistant that does the thinking the method should make people do erodes their skills, one that challenges their reasoning builds them. Our article on AI and problem-solving skills describes both and gives a weekly routine.

Art Smalley frames it as a formula: Impact = Technology × Behavior × Management. Technology is the easiest part to acquire; behavior takes time to build; management decides whether AI produces learning or confusion. Mark Graban, writing on Jeffrey Liker’s Toyota Culture in January 2026, says the same of tools, dashboards, templates “and now AI”: “none of those things substitute for the underlying system.”

Practitioners describe AI as capable but fallible. Gene Kim compares it to summer interns; at the Lean Summit, Grainger’s CTO Jonny LeRoy spoke of an “army of interns” whose failure modes leaders must understand to supervise it, and Eric Dickson, President and CEO of UMass Memorial Health, summed up the combination: “The way to go is AI + A3.”

For a lean leader, the 2026 trends point to one starting point: use AI where lean already sees waste, in the reporting of the management system, with written guardrails, and keep the Check, the Act and the coaching with people. Write prompts that name the method; make the assistant stop and say what is missing; keep a mentor on the reasoning; and tie every AI effort to an objective of the year, so that speed serves the strategy rather than the reverse.

Our guide to AI prompts for lean gives copy-ready prompts for the X-Matrix, the A3, the 8D and the PDCA, written with those rules.