What is AirRating?
AirRating is a B2B flight risk intelligence platform that calculates a Risk Score (0–100) for any flight before booking. It combines historical flight data, machine learning, live weather, and airport congestion signals to quantify the probability of delay or disruption on a specific route, airline, and date — giving corporate travel teams and OTAs a data-backed signal before committing to a booking.How it works
Risk Score 0–100
A single number that summarizes the delay risk for a given flight (route + airline + date).
Multiple signals
1M+ real flight records, live weather (Open-Meteo), airport congestion (Eurocontrol).
AI explanation
Plain-language summary of why a flight is risky, generated by Claude AI.
Scoring formula
- Statistical model: Historical delay rate on the exact route + airline combination
- ML model: XGBoost trained on 500,000 flights, AUC-ROC 0.64, target P(delay > 30 min)
- Confidence weight:
min(1.0, confidence × 2)— automatically reduces ML weight when historical data is sparse
Data coverage
| Source | Detail |
|---|---|
| Flight records | 1,068,253 real flights (Dec 2025 – Mar 2026) |
| Routes | 47,773 aggregated routes |
| Average delay rate | 27.3% |
| Delay threshold | > 30 minutes (IATA standard) |
| European airports | 47 major airports |
| Global hubs | Dubai, New York, Tokyo, Bangkok, Hong Kong, Atlanta, Los Angeles, Istanbul |
Who is it for?
- Corporate travel managers — assess risk before booking business trips
- Online Travel Agencies (OTAs) — surface risk scores at checkout
- Travel Management Companies (TMCs) — integrate into duty-of-care workflows
- Insurers — use risk scores as input for dynamic travel insurance pricing
- Individual travelers — book with confidence using data-backed decisions
Current status
AirRating is in private beta as of March 2026. The API is available to early access partners.The public API is planned post-beta. To request access, join the waitlist at airrating.io.