Developers created an AI-based system for a Russian airline to assist in rescheduling flights during operational disruptions. The system consists of eight specialized AI skills that automate the process of analyzing flight data, passenger bookings, and airport restrictions to suggest cancellations or rescheduling.
The system uses a combination of language models for text analysis and Python scripts for calculations, such as seat availability and flight ranking. It does not connect directly to airline software; instead, employees upload data from scheduling, booking, and BI systems manually.
The developers tested the tool using a historical disruption scenario from July 7, 2026, in Krasnodar, and a major incident on August 18–19, 2024, involving over 230 flights across multiple airports, including Domodedovo and Sheremetyevo.
During testing, the system underwent 164 iterations of the August scenario. The final workflow consists of 13 sequential steps, ranging from parsing disruption notifications to generating a final Excel report for review by commercial staff.
The AI successfully matched the manual decisions made by airline staff in specific test cases, including a scenario where cancelling a pair of flights prevented a two-hour delay for subsequent operations.
