At American Airlines, AI starts with baggage handling
- At a company the size of American Airlines, immediately releasing AI at scale is unrealistic, according to Phillip Easter, head of emerging technology at American Airlines, speaking last week at The AI Summit in San Francisco. Implementing advanced technologies can prove challenging given the legacy technology stack the company has relied on for business operations for more than 60 years. Instead, AA had to think big, but start with small implementations and move quickly.
- The highest return for AI projects is operations, according to Venkata Pilla, manager for machine learning and data science of American Airlines, speaking at the Summit. AA took data from baggage scans to search for patterns and report it back to the business. They found which bags were highly likely to be mishandled, insightful information for the customer and the airline.
- With lots of legacy systems, the goal for AI is not to build a solution that fits into the existing tech framework, according to Pilla. Instead, AA wanted to build a small prototype and web application that could exist outside the standard technology stack. For baggage handling, AA created an application for the station manager which helped prioritize which bag to address first, remediating bottlenecks.
Customers lambast airlines for technology failures, especially since high-profile outages can cause rippling delays. Take Delta's recent technology struggles, for example. Earlier this week, the airline had to temporarily ground flights after a technology issue stopped some systems from operating on Tuesday, according to Delta.
Delta, like other airlines, has started small with its modernization efforts, trying to make the customer experience more fluid.
Part of the modernization goal is differentiating services to customers. "All airlines look the same," said Easter. Just 10% of customers are road warriors, who fly the same airline while they're constantly on the road. The other 90% of customers fly an airline once or less per year.
To the 90%, airlines seats, aircraft and staff all look the same. So that challenge becomes how an airline can leverage cool technology to allow customers to get better services, Easter said. AA can use AI and ML to get a smart employee to focus on solving a problem within their grasp.
Correction: In a previous version of this article, Venkata Pilla was misidentified. He is the manager for machine learning and data science of American Airlines.
Follow Naomi Eide on Twitter