Industrial adhesive control with AI — one drum with out-of-range viscosity means the customer's line is down.

The problem

The recipe is stable, but nobody told the batch.

Any adhesives plant manager knows it: the nominal recipe has been in the ERP for years, but the real batch depends on things the recipe does not capture.

1. Raw material changes from supplier to supplier — the base polymer may come with a slightly different molecular weight, the plasticizer with more moisture than expected, the tackifying resin with a softening point half a degree low. The supplier's analysis is in the LIMS or in a PDF that arrived by email, but it is rarely cross-referenced with the recipe before charging the mixer.
2. The mixer performs differently every day — a worn agitator, a heating jacket losing a couple of degrees, a miscalibrated temperature sensor, a dosing pump that has lost accuracy. Today's viscosity curve is not exactly the one from six months ago, even though the SCADA setpoint has not changed.
3. The customer spec takes no excuses — a drum of EVA hot melt that reaches the customer with viscosity outside the agreed Brookfield range for the application, a reel of laminated paper where the adhesive does not open in time, a PSA that fights back on someone else's thermoformer. Every complaint has a name, a surname and a return-freight cost.

The lab measures at the end of the batch or by sampling the drum, and returns the figure hours later. By then the batch is packed and, with luck, goes to a less demanding customer; with less luck, it comes back. The problem is not the recipe, it is that the classic system is always late — and every complaint from a repeat customer shows up in the whole year's numbers.

How it fits the IRIS system

iLEAN does not change your mixer — it puts the missing intelligence on top of the data you are already generating.

The problem in industrial adhesive control is not a lack of information: it is information living on separate islands — the LIMS knows how the raw material came in, the mixer's SCADA knows what is happening in the jacket right now, the ERP knows which customer is waiting for that drum and at what viscosity, but none of them talks to the next at packing speed. iLEAN acts as the putty that fills those gaps, without asking you to change SCADA, ERP or the packing line.

Edge sees viscosity and temperature in line. Connect reads the LIMS analysis and the customer spec wherever they come from. The agent cross-references them with the expected curve and, if something does not add up, holds the drum before the pallet. The person signs — never the other way round.

The three iLEAN pieces applied to industrial adhesive control:

• Edge — a computer-vision terminal (CNN) with in-line signal capture: in-line Brookfield viscometer, temperature probe, camera over the filling head to read the label and detect bubbles or visual variation in the product. If viscosity moves outside the range accepted for that spec, it triggers a stack light and notifies the operator before the drum is closed. It works with no network.
• Connect — captures the raw-material analysis whether it comes from a modern LIMS, from the lab's spreadsheet or from the R&D manager's own sheet. And it also captures what comes in from outside (a customer spec PDF, an email from the supplier flagging the new resin batch, a WhatsApp message from the logistics supervisor about the shipment) at second zero, with nobody having to forward anything.
• Agent — cross-references the batch recipe, the raw-material analysis, the mixer's actual curve and the customer spec. If there is a deviation, it does not send an email at 10 p.m.: it holds the drum and alerts the quality manager through whatever channel they use. It also generates the traceability dossier — which raw material, how long the mix ran, which shift — ready for an audit or a complaint. The person validates and signs; the pallet never leaves on its own.

See the full IRIS architecture →

Before and after

Manual control vs. cross-referenced control with iLEAN

Aspect	Manual control + LIMS by sampling	With iLEAN Edge + Connect + Agent
Detection of viscosity deviation	When the lab returns the figure (hours)	In line, during packing
Cross-reference with supplier raw material	The operator checks the LIMS if there is time	Connect brings it to the mixer panel automatically
Drum / IBC labeling	Operator reads the label before the pallet	Edge verifies label vs. ERP batch, holds it if it does not match
Traceability for a complaint	Rebuilt by hand from the ERP and from paper	Per-batch dossier, automatic, with a photo of the line
Multi-customer spec (hot melt, PSA, solvent)	Same threshold for every SKU	Model trained per family and per customer
Operation without a network	n/a	Edge keeps running with its own light

Impact estimate

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

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Mid-sized adhesives plant (a mix of hot melt, acrylic PSA and solvent-based), 2-3 mixers and a multi-SKU drum/IBC packing line.
• Edge pilot on one packing line (in-line viscometer + camera over the filling head + integration with the batch recipe in the ERP/MES). First value expected within a few weeks.
• Indicative payback between 4 and 9 months, depending on the frequency of customer complaints and of batches sent for rework documented over recent years.
• The hard lever is a single complaint avoided from a repeat customer: returned drum, return freight, rework, damage to the commercial relationship. The expected reduction in in-line rejects is ≥ 30% against your current baseline.

And the quality director's reasonable doubt

“What if the AI gets it wrong and holds a drum that was fine?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely compares a real measurement against a spec window trained on your own batches, the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN holds and the person signs. The three safety rings are there for exactly this.

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