Aviclaim
An AI claims engine that processes 25,000 cases a year — with lawyers in charge.
Aviclaim recovers compensation for passengers under EU 261. Qubiz built the platform that turns boarding passes and airline correspondence into adjudicated, paid claims — without removing legal judgement from the loop.
The challenge
A volume business with a regulatory backbone — and a team being eaten by paperwork.
Each compensation claim required reading boarding passes, retrieving delay codes from carriers, applying the relevant EU 261 article, drafting demand letters, and tracking responses for years. Aviclaim's legal team was spending most of its time on document handling, not case strategy. Any AI solution had to be defensible in court, traceable to regulation, and aligned with how lawyers actually reason about a case — not a black-box scoring system that no one could explain to a regulator.
Automate the paperwork. Keep the legal reasoning explicit.
Three phases, each validated against a representative legal sample before scaling.
- 01
Document intake
Multilingual extraction across boarding passes, e-tickets, and airline correspondence. The volume cases now route themselves; the edge cases route to a lawyer with full context.
- 02
Adjudication engine
EU 261 and UK CAA rules encoded as a versioned engine. Every output traces from regulation to outcome — defensible in court, auditable to the regulator.
- 03
Carrier orchestration
Direct ingestion from airline systems where APIs exist, monitored mailbox parsing where they don't. The platform now retrieves what the legal team used to chase.
- 04
Continuous improvement
Lawyer feedback closes the loop on the model. The platform improves on the same cadence as case law — not on a vendor release schedule.
WHAT WE BUILT
Document extraction model
A purpose-built extraction model trained on five years of historical flight documentation — boarding passes, e-tickets, airline correspondence in eleven languages. Confidence-scored fields with selective human review only on outliers.
EU 261 / UK 261 adjudication
A versioned legal rules engine encoding EC 261/2004 and UK CAA equivalents. Every adjudication is fully traceable from regulation to outcome, with explicit handling for disputed delay codes and force-majeure exemptions.
Airline data ingestion
Direct ingestion from twenty-plus airline data sources — APIs where available, monitored mailbox parsing where not. Aviclaim's legal team no longer chases delay codes; the platform retrieves them.
Lawyer-in-the-loop
The model surfaces edge cases to legal leads with a structured rationale, not a black-box score. Lawyers spend their time on the cases that actually require legal judgement — not on the 95% that don't.
The outcome
Lawyers doing legal work. Claims clearing on time.
25k
Claims now adjudicated annually with the same legal headcount.
92%
Of cases adjudicated end-to-end with no human routing required.
11
Languages supported across the document extraction pipeline.
0
Adverse regulatory findings on the platform's decision audit trail.
We needed AI that a courtroom could trust, not just one that worked on a demo. Qubiz built it the way our lawyers think, and that's why our lawyers actually use it.
Fleet intelligence that reduced fuel consumption and improved delivery accuracy.
Connected telematics, route planning, and driver behaviour data into a live fleet dashboard with proactive alerting.
Predictive underwriting models that reduced loss ratios by 22%.
Translated actuarial expertise into machine-learning models embedded directly into the underwriting workflow.