Digital Infrastructure
Automation of incident categorisation in AIRLINE
To automate the categorisation of airline reports and automate the end to end process in airline for the initial 3 categories Hazards, Low and grounded, civils.
Key deliverables
- A working proof of concept (i.e. a set of data pipelines incorporating GenAI services) that can extract, transform and categorise in-scope incidents as well as a one-off historical trend analysis of incidents in the last 9 years. This prototype will be developed within UKPN’s tech stack but will be operated in a development environment to allow for rapid iteration and testing. -A final, concise report summarising work undertaken and the findings, including: • Executive summary • Data extraction results and PoC evaluation logic • PoC configuration, finetuning and testing results • Historical trend analysis results • Indicative cost analysis associated with different pipeline configurations and expected UKPN call volumes • A technical integration plan and recommendations on next steps to enable the PoC to be integrated with UKPN’s systems and deployed in production
Target outcomes
- For AI to categorise the 3 categories in Arline and provide a trend analysis
Measures of success
- It removes some of the manual work the team are carrying out with the aim after the proof of concept is delivered to automate completion of fields in Airline and other sites for example the HSE site for Reg
Stakeholder Engagement
Tony Cohen, Kai Patel, Jenny Cogan, Nural Islam, Martin Knoebel, E & Y who are carrying out the work