Milan's AI reads the complaint before a human answers
SEA Milan Airports' use of Azure OpenAI for customer correspondence is a defined back-office application. A reported reduction in response time is attributed to project participants, while a biodiversity robot remains a trial.
Control tower and Terminal 1 at Milan Malpensa Airport. Photograph: Arne Müseler / CC BY-SA 3.0 de (cropped to fit). Shows the airport, not an AI implementation.
The picture at Milan Malpensa
The operating setting
Customer letters arrive in varied forms and can be difficult to classify consistently. SEA Milan Airports uses AI to identify topics, priority and sentiment and prepare summaries for its customer-service workflow.
The work is less visible than terminal robots, but has a clear owner and task. The later OLGA biodiversity-monitoring test should not be portrayed as another proven production result. An automated category still needs human correction when it is wrong. If a message about accessibility or a missed connection is assigned the wrong priority, the employee must recognise that mistake before the model's summary becomes the basis of the reply.
Selected implementations
The application and the limits of the evidence
01 / Deployed
Milan's AI reads the complaint before a human answers
The operational problem
Airport support staff need to triage enquiries correctly and reply before cases become stale.
What the sources report
SEA Milan Airports and Arsenalia described Azure OpenAI tagging and summarising tickets in 2025, with a reported 30% improvement in email response time.
Why it matters
The task can be audited using routing accuracy and handling time, rather than general claims about generative AI.
What remains unproven
The 30% figure is project-reported; baseline, human correction rates and independent verification are not published.
Evidence supports a reported claim; it does not automatically establish a measured operating benefit.
The wider register
What else is happening at Milan Malpensa?
The wider register below keeps any other recorded technology separate from this selected operational story. Its source and status should be assessed independently.
No additional publicly listed deployment records.
Aerostack assessment
A useful follow-up would describe how often the system misclassifies an enquiry and what staff do when it does.
The published sources remain available alongside the underlying implementation record; claimed results should retain their original attribution.
Live data from the Aerostack register. Status reflects published evidence only. See Our approach for how records are graded.
Aerostack register · 1 tracked deployment
Aerostack tracks AI deployments across airports and airlines, with the published evidence behind each record. Photography is real aviation imagery. Record thumbnails may show the named airport or airline; other photographs do not.
