CBAM with AI — the European importer wants the real number, not the flattering average.

The problem

A footprint per batch is three pieces of data that are almost never visible at the same time.

An honest CBAM report is the cross-check of three realities that live on different islands:

1. Real energy consumption of the batch — kWh, m³ of gas, steam, on the specific line and in the specific window in which it was made. It usually sits in the SCADA, in the utility meter, in a PLC nobody ever integrated.
2. Emission factor of the electricity grid — it varies by hour of the day. That same tonne of steel, made at night with wind power or at midday with coal, does not carry the same footprint.
3. Supplier emission factor — for the scrap, the lime, the clinker, the ammonia. It arrives as a PDF by email, in a spreadsheet, sometimes through a supplier portal — and almost never with a clear version number or in time for clearance.

If one is missing, the European importer applies the default factor set by the Regulation, which penalises. And if you hand everything over aggregated into a flattering annual average, the independent verifier gets suspicious — rightly so. A defensible CBAM file is not a figure: it is a chain of signed data, per batch, ready to audit.

How it fits into the IRIS system

iLEAN does not add another sheet to the report — it seals the cracks between the systems you already have.

The CBAM problem is not a lack of information: it is information living on islands that, at the critical moment (a shipment with a tight verifier deadline, a supplier who does not deliver the factor in time), never reaches the person who signs. iLEAN acts as the filler that seals those gaps, without asking you to change ERP, SCADA, or your relationship with your supplier.

Connect captures the meter, the supplier email and the hourly grid curve. The agent cross-checks against the batch and the customs destination. If a critical piece of data is missing, it holds the shipment before clearance. The person signs — never the other way round.

The three iLEAN pieces applied to CBAM:

• Edge — captures on the line the consumption of the specific line producing the batch: electricity meters, gas consumption, steam. It works without a network: the consumption data is recorded against the batch even if the utility meter has an incident.
• Connect — listens to the supplier email, the portal where the certificate is uploaded and the hourly API of the electricity grid (grid mix per window). A certificate arrives as a PDF and at second zero it is converted into JSON with its version number. And it captures outward too: the importer's note with the verifier's new rule, the buyer's message with the changed Incoterm.
• CBAM agent — lives in Central, inside the three security rings. It cross-checks the batch, the consumption, the hourly grid factor, the supplier factor and the customs declaration. If everything adds up, it prepares the file with a timestamp. If a piece is missing, it holds the shipment and alerts the owner. The person validates and signs; the critical operation never runs on its own.

See the full IRIS architecture →

Before and after

CBAM reporting in a spreadsheet vs. a per-batch file with iLEAN

Aspect	Spreadsheet + emails + default factor	With iLEAN Connect + CBAM agent
Data granularity	Annual or quarterly average	Per batch number, with a timestamp
Electricity grid factor	Annual, declared by the utility	Real hourly curve for the manufacturing window
Supplier certificate	PDF in the purchasing inbox, with nobody reading it in time	Captured at second zero, signed JSON, versioned
Defence before the verifier	«Trust our average»	Chain of signed data per batch, auditable
Risk of the default charge	High — it penalises whenever a piece is missing	Low — the agent alerts before clearance
Quarterly CBAM report	Weeks of manual work	Generated by the agent; the person reviews and signs

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.

• Steel mill, aluminium plant, cement works, fertilisers or hydrogen with relevant EU exports. Today it reports with the default factor or with an annual average.
• Connect + CBAM agent pilot: meter capture + grid mix API integration + inbox for critical supplier emission factors + verifier dashboard. First value expected within a few weeks: the first batch with a signed file.
• Indicative payback between 4 and 9 months, depending on the share of revenue exposed to CBAM and the gap between your real factor and the default factor set by the Regulation.
• The hard lever is removing the default charge on the clearances where your real data beats the penalising one, and cutting the internal cost of assembling the quarterly report.

And the sustainability lead's reasonable doubt

«What if the AI hallucinates an emission factor?» — hallucination is a problem of free generation, not of anchored tasks. In tasks where AI only recontextualises a piece of data from one system into another (reading the supplier certificate and converting it into JSON, cross-checking the batch consumption against the hourly grid curve), the best models brought the error below 1.5% ^[1]. And even so, the critical part is never decided alone: the file is not sent to the importer without a person signing it. The three security rings exist precisely for this.

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