Digital PDCA on a 30-day cycle — because the Check always dies in a forgotten spreadsheet.

The problem

PDCA does not die in the Plan. It dies in the Check nobody runs.

Any black belt or continuous improvement manager with more than a year on the floor has seen it. The team opens a PDCA full of energy: a nice A3 sheet, a clear metric, an assigned owner. Three weeks go by. Check day arrives and the owner is putting out a fire on line 7. Nobody pulled the metric out of the MES. The A3 hangs on the shop-floor board for another two months and nobody mentions it again. The audit comes and it is closed as “implemented”, when what was implemented was hanging it up.

1. Plan — the easy part. There is time, there is a meeting, there is a team photo.
2. Do — it gets executed, more or less.
3. Check — this is where most of them die. It requires pulling the data from the MES, cleaning it, comparing it against the baseline, discarding the noise. Unglamorous work that always gets postponed.
4. Act — without a real Check, the Act is an opinion. And opinion does not standardize.

The classic setup (a sheet, an Outlook reminder, a monthly improvement meeting) works for the first cycle, sometimes the second. By the fourth it does not. And the PDCA culture that is supposed to be the competitive edge of Lean stays on the poster on the wall.

How it fits the IRIS system

iLEAN does not add a fourth system — it seals the crack between the A3, the MES and the calendar.

PDCA does not fail for lack of method. It fails because of data living in islands and an owner who cannot be everywhere. iLEAN acts as the filler that covers that dead zone between the shop-floor board, the MES and the manager's calendar, without asking you to change your A3 template or your monthly ritual.

The Writer drafts the A3 on the template you already use. The Agent hooks the metric to the real data and after 30 days delivers the Check already done. The person signs the Act — never the other way around.

The two iLEAN pieces applied to the 30-day PDCA:

• iLEAN Writer — when the Plan is closed, it generates the A3 on the template your organization already has. It does not reinvent the format; it fills in the fields the system knows (metric baseline, last twelve months of history, similar PDCA closed in other plants) and leaves the ones requiring judgement to you.
• iLEAN Agent — it lives in the Central. After 30 days it comes off the calendar and runs the Check: it pulls the metric from the MES or the SCADA or the spreadsheet the shift supervisor updates every morning, compares it against the baseline, discards statistical noise, writes up the result and proposes the Act. If the Plan set the target at −20% scrap and it landed at −9%, the agent does not dress up the number — it proposes opening a second cycle and says why.

The improvement director sees a dashboard with the live PDCA across every plant, the cycle day each one is on, and the accumulated rate of closures with an actionable conclusion. This is what we call the warm bed: the shift walks in and finds the cycle closed, ready to decide on.

See the full IRIS architecture →

Before and after

PDCA on a sheet vs. a 30-day PDCA with iLEAN

Aspect	Classic PDCA on a sheet	With iLEAN Writer + Agent
Metric baseline	Calculated by hand, sometimes not calculated at all	12 months of history read from the MES at second zero
Check reminder	An Outlook entry the owner postpones for a real emergency	The agent runs the Check without asking the owner
Source of the measurement data	The shift supervisor's spreadsheet, copied by hand	MES / SCADA / IoT / spreadsheet read by Connect, untouched
Act decision	A meeting postponed to next month	Three proposals ready to sign, based on the real result
Knowledge across plants	Every plant repeats the same PDCA	The corporate dashboard cross-references similar PDCA and proposes consolidating
Share of PDCA closed with an actionable conclusion	20-40% typical (estimate to be validated)	Above 70% within a few cycles (estimate to be validated)

Impact estimate

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

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

• Industrial operation with 1-3 plants and a continuous improvement programme opening 8-15 PDCA per plant per year. Today between 20% and 40% close with an actionable conclusion (estimate to be validated).
• Deployment of iLEAN Agent + Writer on top of the programme calendar. First value within the first 30-day cycle: the first Check comes out on its own and the team notices.
• Indicative payback between 4 and 9 months, depending on the cost of the black belt time freed up and the savings from each PDCA that now really does close with a real conclusion.
• The hard lever: short cycles = more cycles per year = compounding improvement. A 30-day PDCA that genuinely closes is worth more than three 90-day PDCA that fizzle out.

And the improvement manager's reasonable doubt

“Won't the agent get the Check wrong?” — hallucination is a problem of free generation, not of anchored tasks. Pulling a metric from the MES, comparing it against a baseline and applying a statistical test is the anchored task par excellence: on this kind of task, the best models brought the error below 1.5% ^[1]. And even so, nothing critical is decided alone: the agent proposes the Act and the person signs. The three safety rings exist precisely for this.

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