AI suppository manufacturing control — one degree less in the mass is a run of weights out of spec.

The problem

Three plant realities that almost never get cross-checked in time.

On a suppository line — cocoa butter, Witepsol or gelatin-glycerin base, dosed into a strip — the guarantee that units are within specification rests on three pieces of data that in most plants each live on their own island:

1. The temperature of the mass and of the mold — the key to density, shrinkage and final weight. Sometimes in the PLC, sometimes only in a thermocouple wired to an old local panel, sometimes in a spreadsheet kept by the formulation lead.
2. The dosing into each cavity of the strip — what volume the machine is calling for and how it is varying from cavity to cavity. Verified by IPC sampling, not continuously.
3. The thermal sealing of the strip — head temperature, pressure, speed. The one thing the operator checks by eye on a sample at the start of the batch.

The quality manager knows everything adds up for 99% of the batch. That 1% is what triggers the OOS investigation into weight, the suspicion over strip integrity, and the hold on the whole batch. Caught late it is a deviation; caught in line, it is nothing.

How it fits the IRIS system

iLEAN does not add a fourth system — it seals the cracks between the three you already have.

The problem in suppository manufacturing is not a lack of information: it is information living on islands that, at the critical moment (one degree less in the mass, a change of excipient batch, a sealing head drifting down through wear), does not reach the decision-maker in time. iLEAN acts as the putty that fills those gaps, without asking you to change your Sarong/IMA/Servac filler.

Edge sees the filled cavity and the sealed strip before cartoning. Connect reads the temperature of the mass wherever it lives — PLC, old thermocouple, spreadsheet. The agent cross-checks against the recipe and the GMP specification and, if something does not add up, holds the run. The person signs — never the other way round.

The three iLEAN pieces applied to suppository manufacturing:

• Edge — a terminal with machine vision (CNN) over the strip outfeed. It verifies the fill of every cavity, the presence of internal voids from shrinkage and the thermal seal line in milliseconds. It fires the actuator (strip rejector, stack light, selective stop) before cartoning. It works without a network. If the plant loses Internet, Edge keeps reading and holding.
• Connect — captures the filler whether the data comes from the PLC over OPC-UA, from the old terminal by OCR, or from the formulation lead's setpoint spreadsheet. It also captures the temperature of the bath, of the mold and of the sealing head, and whatever arrives from outside (an email from the R&D lead with a change to the formulation, a WhatsApp from the supplier about a deviation in the excipient batch).
• Agent — cross-checks mass temperature, dose per cavity, strip sealing, the MES recipe and the GMP specification. If there is a deviation, it does not send an email at 10 p.m.: it holds the run and alerts the quality manager through whatever channel they use. The person validates and signs; the line does not restart on its own.

See the full IRIS architecture →

Before and after

Suppositories controlled by manual IPC vs. cross-checked with iLEAN

Aspect	Manual control + IPC by sampling	With iLEAN Edge + Connect + Agent
Weight verification	Periodic sample on a balance, assume the rest	Every cavity verified in line, in milliseconds
Mass temperature	Measured at batch start, assumed stable	Correlated with the dose cavity by cavity
Void from shrinkage	Detected if the operator breaks the sample open	CNN vision before cartoning, selective rejector
Thermal sealing of the strip	Checked by eye on a sample at the start	Weld line verified strip by strip
Operation without a network	n/a	Edge keeps running on the panel's own power
File for the FDA/EMA auditor	Rebuild from paper, days	Automatic per-batch dossier, with images and traces

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 pharma plant, one line of pediatric and adult suppositories (Sarong / IMA / Servac), 80-150 strips per minute, multi-format with mold changeovers.
• Edge pilot on one line (cameras over the strip outfeed + actuator + integration with the filler and thermocouples + MES). First value expected within a few weeks: detection of under-filled cavities / poor sealing above the manual standard.
• Reduction of OOS deviations from weight or strip integrity ≥ 30% in the first pilot batch, depending on initial severity and filler dispersion.
• Indicative payback between 4 and 9 months, depending on the frequency of documented incidents and the average cost of rework/destruction.
• The hard lever is a single OOS deviation avoided: investigation, rework, risk of a scrapped batch, the auditor's attention.

And the quality manager's reasonable doubt

“What if the AI gets it wrong and lets a strip with a bad cavity through?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely recontextualizes a piece of data from one system to another (reading the cavity and comparing it with the setpoint), the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN holds the strip 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.

← AI syrup filling control · AI nasal spray control → · AI pharma blister control →