mRNA vaccine batch in a -70 °C chain — an unrecorded break is not an incident, it is an invalidated batch.

The problem

The -70 °C chain is real — traceability of its breaks, almost never.

In a biotech plant that manufactures or stores mRNA vaccine, the -70 °C ultra-cold chain is the product's red line. But the record of what happens between the freezer sensor and the batch dossier still runs with three classic gaps:

1. Loggers on islands — each ultra-low freezer spits out its own CSV or connects to its own software. Nobody cross-checks the curve against the batch that is inside it.
2. Invisible excursions — the door left open for half an hour by night maintenance stays in the log, but not in the batch file. By the time a quality control check catches it, traceability has to be rebuilt by hand.
3. Missing owner — the log says “excursion at 03:42”, but it does not know who was there, what they did, or whether they acted within the SLA.

The QA manager knows it. The maintenance team knows it. But between the sensor and the dossier there are three systems, two shifts and a lot of goodwill — and in an EMA or FDA audit, that is not enough. One mRNA batch invalidated through insufficient traceability costs far more than years of licence for any traceability system.

How it fits the IRIS system

iLEAN Tracer does not replace your loggers — it seals the gap between the sensor, the person and the dossier.

Ultra-cold traceability is not an instrumentation problem: it is a problem of cross-checking the sensor curve, the batch order in the MES, the badge of whoever was in the room and the SLA that applies. iLEAN Tracer is the piece of the IRIS category that seals that gap — without asking you to change the ultra-low freezers, the loggers or the MES. The filler, not the wall.

Edge sees the room. Connect reads logger, badge and MES wherever they live. The agent composes the signed genealogy of the batch. The person validates the deviation — the system does not close it on its own.

How the work is shared between the iLEAN pieces for this case:

• Edge — a terminal in the room cabinet that reads the ultra-low freezer logger locally. It works without a network: if the plant loses WiFi, Edge keeps signing events locally and syncs them when it comes back. What is critical does not depend on connectivity.
• Connect — captures the room access badge (who came in, when they left), the active batch order from the pharma MES, the maintenance plan (whether the door was open for an authorised intervention) and the applicable SLAs. It also captures what arrives from outside (an email from the manufacturer with a change to the product data sheet) at second zero.
• Tracer agent — composes the signed event per break: timestamp, value, duration, identified owner, applicable SLA and reason. If the SLA is missed, it alerts the QA manager through whatever channel they use and leaves the batch “under review”. The person validates whether it continues or is invalidated — the system does not sign the release on their behalf.

See the full IRIS architecture →

Before and after

Manual ultra-cold traceability vs. signed traceability with iLEAN Tracer

Aspect	Logger + maintenance log	With iLEAN Tracer
Sensor reading	One CSV per freezer, on islands	Unified and signed at second zero
Identity of the owner	Manual note by the shift	Room access badge, cross-checked automatically
Response SLA	Implicit in the procedure	Explicit, measured and signed into the event
Planned vs. real excursion	Rebuilt by hand	Cross-checked against the maintenance plan
Batch dossier for EMA/FDA	Rebuilt over weeks	Self-contained, ready in hours
Operation without a network	n/a (standalone logger, manual cross-check)	Edge keeps signing locally, syncs on return

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 from your plant. We lay it out so the committee has an order of magnitude; we refine it during the diagnostic.

• Mid-sized biotech plant with 4-10 ultra-low freezers at -70 °C, a pharma MES and badge access control already running.
• Tracer pilot in one critical room (Edge in the cabinet + integration with the loggers + integration with MES and badge). First value expected within a few weeks: the first signed event with an identified owner.
• Indicative payback between 4 and 9 months, depending on the documented frequency of batches under review and the average cost of invalidation or reprocessing.
• Expected reduction in the time needed to prepare the batch dossier for audit: ≥ 30%. The hard lever is a single invalidated mRNA batch avoided — it pays for the pilot and several years of licence.

And the QA manager's reasonable doubt

“What if the AI adds noise to the log and the audit throws it back at me?” — hallucination is a problem of free generation, not of anchored tasks. Composing an event from sensor readings + badge + MES is exactly an anchored task, where the best models brought the error below 1.5% ^[1]. And even so, the critical decision is not signed on its own: Tracer leaves the batch “under review” and the person signs. The three iLEAN security rings are designed so that the OT network only accepts what has been validated, packaged, signed and approved by a person.

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