Azzera, a Montréal, Canada-based provider of aviation emissions compliance and carbon markets software for business aviation, is developing an AI-powered verification layer for its Celeste platform that automatically cross-checks an operator's flight data against standard authoritative external sources, including EuroControl, NAV Canada and CRCO, to catch discrepancies while preparing reports.

According to the company, aviation operators managing emissions obligations across the EU ETS, UK ETS, Swiss ETS, CORSIA and ReFuelEU often do not discover a data discrepancy, a flight missing from one system, a mismatched fuel figure, until an auditor finds it first. The compliance burden keeps growing. ReFuelEU's SAF mandate rises from two percent in 2025 to six per cent by 2030, and CORSIA turns mandatory for most member states in 2027.

“Operators do not find out about a data problem until the last minute, and by then it's an administrative nightmare to iterate through the datasets,” says Puja Mahajan, CEO and founder of Azzera. “We can't change when the external data sources become available. However, we can help operators respond when discrepancies surface late in the process.”

The system identifies flights present in one system but missing from the other and flags mismatches in flight level data between an operator's own records and independent sources. Where earlier this kind of gap surfaced only during manual reconciliation or a verifier's review, the AI layer generates a plain language explanation for why a discrepancy occurred, turning what used to take hours of investigation into an immediate, actionable answer for the team reviewing it.

“Every operator we work with is trying to reduce their time-to-audit; comparing data quickly and with confidence is the way to get there,” explains Anant Jain, chief operating officer at Azzera.

The system treats data confidentiality as a non-negotiable. The verification layer runs inside a contained process built only to compare an operator's data against approved sources. When a discrepancy needs a human reviewer, that review sharpens how Azzera supports future cases; it does not train the underlying AI model. Client data stays inside each operator's own Celeste account, is never pooled across clients and is never used to fine-tune, train or benchmark the model itself, a standard applied across Azzera's platform, including its SAF POD beta programme. Compliance data carries real financial and legal weight, and Azzera treats that boundary as non-negotiable, positioning itself among the leaders in safe AI development for aviation decarbonisation.

Azzera is receiving advisory services and funding from the National Research Council of Canada Industrial Research Assistance Program (NRC IRAP) to enable a research and development project focused on flight data verification and compliance reporting. The project extends Azzera's history of federally supported innovation, including prior funding from Sustainable Development Technology Canada, Innovative Solutions Canada, along with its ongoing INSAT-funded SAF POD initiative.

The verification layer is now testing with existing Celeste clients against its first set of external data sources, with coverage expanding over time, ahead of a full rollout across Azzera's base of more than 50 aircraft fleet operators in Europe and North America.