The Vintage Press Panel That Starts Talking

The problem

The press that stamps your critical body panel forgets every stroke.

The vintage press is usually the bottleneck for a critical part, and its force/alarm data dies on screen every few seconds because nobody logs it stroke by stroke. Without that correlation, a scrap increase on a specific part can't be attributed to a stamping process drift until the problem is already large.

• Most OEM stamping shops keep at least one older-generation press running a critical outer panel — a hood, a door outer, a body side. Its proprietary control sits off the network by design.
• The force curve and the alarms show on the panel for a few seconds and are gone. Nobody logs them stroke by stroke, because nobody can stand in front of a screen all shift.
• So when splits or necking start showing up on that panel in the body shop, nobody can tell whether the press drifted, the blank batch changed or the lubrication did.
• The question gets answered when the scrap pile is already large, and that panel is often the one holding the whole body line hostage.

How it fits the IRIS system

Connect in photo mode on the press control — read from the outside.

Connect photo mode over the panel: a fixed industrial camera captures the proprietary control's screen; OCR plus a grounded LLM interpret each frame, structure the force curve by batch/part, and cross-check which work order was in the press at that moment.

The press stays exactly as it is: same control, same software, no new port. What changes is that every stroke leaves a curve behind, tied to the work order and the blank batch that was in the press at that moment. On the machine nobody dares to touch, that is the only approach that gets approved, and it is installed in a single shift.

See the full IRIS architecture →

Before and after

The stamping press today versus the stamping press being read

Aspect	Today	With iLEAN Connect
Force curve per stroke	Seen for seconds on the panel	Stored and tied to the work order
Split or necked body panel	Counted as scrap	Traced to stroke, shift and blank batch
Alarms on the old control	Reset and forgotten	Logged with timestamp and context
Process drift on the outer panel	Noticed when scrap is already high	Visible as a trend before it bites
The proprietary press control	—	Untouched, no network opened
Press replacement CAPEX	On next year's plan	Deferrable with evidence

Zero correlation between press force and part cracking/breakage → batch-level root-cause analysis in minutes.

Impact estimate

Impact estimate — to be validated with your numbers.

The block below is an estimate to be validated against your plant's actual data. We put it forward so the committee has an order of magnitude; we refine it during the assessment.

• Estimated payback 5-10 months, before counting the replacement investment it helps defer.
• Scrap reduction on the critical panel once press force can finally be correlated with splits and necking.
• Avoided press-replacement CAPEX: the press gets replaced when the data says so, not because it is blind.
• And batch-level root cause in minutes instead of a week of hypotheses traded between stamping and the body shop.

Estimated payback 5-10 months, plus avoided press-replacement CAPEX. *Figure to be validated*. (Payback 5-10 months · avoids press replacement CAPEX)

And the fair question from the production manager

“Can a camera on an old panel be trusted for a root-cause analysis?” — the task is anchored: a fixed screen, values in known positions and physical ranges the press cannot exceed, where the best models drop below 1.5% error [1]. And a frame that reads a force outside the press's possible range is discarded rather than stored, so the series you analyze is one you can defend in front of the body shop and quality. The read can also be checked against the press counter, which never goes backwards.

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