A3 Thinking digital with AI — a good A3 takes days, not because of thinking, but because of filling in.

The problem

The A3 doesn't die because it's bad — it dies because filling it in is more tiring than the conversation.

Any lean coach with a few years of rolling out A3s has felt it. The methodology is elegant and it teaches people to think — the problem on the left half, the solution on the right half, all on one sheet. But when the A3 owner sits down to write it, they face two hours in the MES pulling the metric history, another hour drawing the Ishikawa, half an hour more checking whether anyone did a similar A3 last year. By the time the meeting with the coach comes, the owner is exhausted and the coach finds a half-empty A3 with stale data.

1. Current situation and problem — demands the MES figure, the photo of the defect, the history of occurrences.
2. Analysis (5 Whys / Ishikawa) — the heart of A3 thinking. But if you start from dirty data, the hypotheses come out crooked.
3. Plan — the part where the coach really adds value, and the one they almost never reach with energy left.
4. Follow-up — the KPIs nobody ends up watching until it's too late.

The classic system (paper template + weekly meeting with the coach) works for the first A3, sometimes for the first three. The tenth one of the year, no longer. And the A3 culture — the one that in theory is the lungs of lean — ends up in a SharePoint folder full of half-filled A3s.

How it fits the IRIS system

iLEAN doesn't think the A3 for you — it removes the plumbing so you can think.

The A3 is not a document, it is a thinking method. But that method suffocates when the data lives in islands and the A3 owner spends 80% of the time collecting and 20% thinking. iLEAN acts as the filler that completes the upper-left half of the A3 with real data at second zero, leaving your energy for the analysis and the plan.

Writer fills the left half with plant data. Agent proposes a first draft of the Ishikawa and the 5 Whys, based on previous similar A3s. The coach and the owner think — never the other way round.

The two iLEAN pieces applied to A3 Thinking:

• iLEAN Writer — works on the A3 template your organization already has. It fills in the fields the system knows (baseline, history, occurrences, photos of the problem captured by Connect) and leaves blank the ones that require judgment (what is the real root cause?, what countermeasure do we plan?). It doesn't reinvent the format; it removes the friction.
• iLEAN Agent — proposes a first draft of the root-cause analysis by cross-referencing similar A3s closed at other plants, the MES history and the veteran's knowledge captured by Connect in immersion sessions. The agent proposes — the team discusses and signs. If the coach sees that the proposed Ishikawa has the wrong bone, they change it; that is exactly the A3 thinking the method is after.

And for the corporate coach: a board with every live A3 across every plant, the phase each one is in, the ones stalled for more than two weeks — and the agents proposing yokoten (connecting A3s with similar problems at different plants) without anyone having to ask.

See the full IRIS architecture →

Before and after

A3 on a blank template vs. A3 with iLEAN Writer + Agent

Aspect	Classic A3	With iLEAN Writer + Agent
Current-situation data	The owner pulls it by hand from the MES	Writer brings it at second zero from MES/SCADA/spreadsheet
History of occurrences of the problem	The veteran's memory	Automatic cross-reference with A3s closed at other plants
Ishikawa / 5 Whys draft	Blank page	Draft proposed from similar A3s — for the team to challenge
A3 preparation time	6-10 hours of the owner's time (estimate to be validated)	1-2 hours to discuss the draft (estimate to be validated)
KPI follow-up	A calendar reminder that gets forgotten	Agent fires an alert to the owner when the metric moves
Yokoten across plants	"We'll tell that story at the next convention"	Agent proposes to the coach connecting owners with similar problems

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the actual data of your program. We lay it out so the committee has an order of magnitude; we refine it during the diagnostic.

• Operation with 1-4 plants, lean program with the A3 as a mandatory artifact of any kaizen or problem-solving effort. Today between 30 and 80 A3s are opened per plant per year, with an average preparation time in the order of 6-10 hours per A3 (estimate to be validated).
• Rollout of iLEAN Writer + Agent on top of the corporate A3 template. First value within a few weeks: the first A3 comes out with its left half filled in minutes instead of hours.
• Indicative payback between 4 and 9 months, based on owner and coach hours freed up and on A3s that reach the Plan phase with enough energy left for a real solution.
• The hard lever: more A3s that reach Plan and Follow-up. An A3 that dies halfway serves nobody; one closed with a measured improvement is worth ten.

And the lean coach's fair objection

"If the AI proposes the Ishikawa, won't people lose the ability to think?" — the opposite. The A3 dies when filling it in is more tiring than thinking it through. Today the owner burns out collecting data and arrives at the analysis empty. With the draft made by the agent, the team discusses the proposed Ishikawa and challenges it, which is precisely the exercise the method is after. Hallucination is a problem of free generation, not of anchored tasks — the best models pushed the error rate below 1.5% ^[1] on tasks like this one. The person supplies the judgment; the line never restarts on its own.

^[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.