Description

1. Product Overview

CarbonBorder AI is a compliance workspace for companies importing carbon-intensive goods into the European Union. It turns two normally disconnected data sets — the customs declarations that record what physically crossed the border, and the supplier evidence that describes how those goods were produced — into a single, traceable carbon border position: what is in scope, what it is likely to cost in certificates, and what evidence supports that figure.

The product is built for the definitive phase of the EU Carbon Border Adjustment Mechanism (CBAM), which applies to imports from 1 January 2026. From that date importers above the 50-tonne annual threshold carry a certificate liability that must be declared and settled for the first time on 30 September 2027, with certificate sales opening on 1 February 2027 (Regulation (EU) 2023/956; Council position on the 50-tonne de minimis). The obligation is financial and it is dated. The number behind it, however, depends on evidence that most importers do not yet hold in a usable form.

CarbonBorder AI is therefore positioned less as reporting software and more as the system of record for a number that finance, procurement and the compliance function all have to defend.

2. The Business Problem

CBAM moves an importer's embedded emissions from a sustainability reporting exercise into a cost of goods. Three problems follow, and all three are operational rather than conceptual.

The evidence lives outside the systems that hold the liability. Customs brokers deliver declarations; suppliers deliver mill certificates, activity data and utility bills in inconsistent formats; procurement holds contracts but rarely holds production data. The importer is legally responsible for the declared figures while depending on parties who have no direct obligation to supply them in a usable form.

Falling back to default values is expensive by design. Commission default values are conservative and carry a statutory mark-up. They are a legitimate choice, but they are a cost decision, and for complex goods a large share of embedded emissions sits upstream at the precursor stage — meaning even a cooperative first-tier supplier does not by itself protect the importer from default exposure.

The declaration must survive verification. Actual values have to be traceable to source documents. Spreadsheets and email threads rarely preserve that lineage across a year of imports, product changes and personnel turnover, and a figure that cannot be traced back is a figure that cannot be defended.

The consequences are measurable in the customer's own currency: certificate spend set by the weakest link in the evidence chain, reconciliation effort concentrated into a single unforgiving deadline, and a position that cannot be explained to a verifier, an auditor or a board without a manual rebuild.

3. How the Product Creates a Solution

CarbonBorder AI replaces the annual reconstruction with a continuous record. Imports are classified as they arrive, against the goods scope, and accumulated per legal entity so that the 50-tonne threshold is answered with data rather than assumption — including the case where a line cannot be classified, which is reported as an open item rather than quietly counted as in scope.

Evidence collection is moved to the supplier. A named supplier receives a single-purpose link, submits installation and activity data, and the document is registered with its own fingerprint the moment it arrives. Extracted values carry their source, their confidence and a review trail; when a human corrects a value, the correction supersedes the original without destroying it. A supplier who cannot provide data becomes a visible, quantified gap — not a silent default.

Calculation is deliberately mechanical and repeatable. Each production line is computed from stored inputs and a versioned set of regulatory parameters, and each run records the basis on which it stands: full methodology where the data supports it, an explicitly labelled upper bound where a required input is missing, or no result at all where there is nothing to compute from. Recalculating a line supersedes the earlier run rather than overwriting it, so the history of a number is as durable as the number itself.

Assessment and sign-off sit on top. Approved runs are aggregated into a compliance position that states the certificate estimate and the basis of that estimate in the same sentence, records who approved it, and can be locked once an accredited verifier's attestation covers the underlying data set. Alongside it, quarterly exposure planning translates the position into the holding pattern the regime requires, using published price parameters rather than internal guesses.

Two refusals are part of the design and are visible in the interface. The product does not invent free-allocation benchmarks it does not hold: it states the position as an upper bound until those inputs exist. And it does not pretend to file: declaration output is marked as a draft, not as schema-validated submission, because the platform has not been verified against the customs authority schema. Buyers are told what they are buying.

4. Problem-to-Outcome Flowchart

5. Value Creation Model

6. Key Business Benefits

7. Before vs. After

8. Customer Workflow

Adoption is designed to follow the importer's own calendar rather than a migration project. A compliance lead connects the current year's import population, enrols the highest-exposure suppliers first, and works the queue of open items; finance consumes the position and the quarterly plan; the verifier receives a locked, attestable data set. Each step produces something usable on its own, which is what makes the first month valuable before the first declaration.

9. Who Benefits

10. Business Impact

Illustrative only: an importer moving a portion of its volume from default to verified values reduces the marked-up part of its liability in proportion to the gap between them. The size of that gap is a function of the supplier's actual performance and cannot be assumed in advance.

11. Product Differentiation

12. Strategic Value

CBAM is the first carbon mechanism that behaves like a customs duty, and it is being widened: additional downstream steel- and aluminium-intensive products have been proposed for inclusion from 2028. Importers who build an evidence system now are not only answering one deadline; they are creating the capability to absorb a broader scope without restarting the exercise.

That capability has value beyond compliance. It standardises how emissions data is requested, received and approved across business units; it gives procurement an emissions dimension in sourcing decisions; and it produces an auditable position that can be reported upward with the same confidence as a financial figure. Because the platform is explicit about what it does not yet know, it can grow with the regime rather than needing to be replaced when the rules or the benchmarks arrive.

13. Closing Value Proposition

By 2027 every importer above the threshold will have to state a carbon number and settle it. The difference between a good and a poor outcome will not be the arithmetic; it will be whether the evidence behind the number exists, is traceable, and is honest about its own limits. CarbonBorder AI exists to make that the normal state of the record — for compliance teams who must defend the figure, for finance teams who must fund it, and for the suppliers whose data decides it.