In-flight Wi-Fi, like smartphones and social media, is a convenience that’s become tough to imagine doing without.

A pioneer in this field is Gogo, which piloted its first in-flight broadband system 10 years ago and now services over 17 of the world’s largest airlines, including British Airways, American Airlines, Air Canada, Alaska Airlines, and Delta. You might not associate artificial intelligence (AI) with airline broadband, but it’s one of the key ways Gogo ensures that its services — which include streaming television and movies, live television, and text messaging — run smoothly and efficiently.

“A lot of our focus has been on getting all [of the data we’re collecting] in one place and enabling analytics,” John Wicklein, senior director of information management, said at VentureBeat’s Transform 2018 conference. “You have a large number of data sources, and you need to understand the health of the aircraft as it flies across the sky.”

In that respect, Gogo has a veritable firehose to work with: the more than 2,500 commercial aircraft and 6,600 business aircraft equipped with its Wi-Fi services.

According to Wicklein, Gogo’s AI journey started with structured data — variables like system uptime, latency, and overall quality of service. Next came unstructured data, like the number of concurrent Wi-Fi sessions and video views during a flight.

He gave one example: the log files from servers stored in aircraft. Gogo records the machines’ fan speeds, which are an indication of their temperature.

“We offload it using a cellular network [when the plane’s on the tarmac] and perform analysis on it,” Wicklein said.

Those data and AI-driven inquiries inform day-to-day maintenance — they’re used to build decision trees for Gogo’s server operators on the ground and help predict equipment failures.

“We look at the results and, in instances where we pulled [a server] box and it turned out that nothing was wrong, [feed that information] into the network,” he said.

Equipment monitoring isn’t the only thing the company is targeting with AI. It’s tapping a suite of AWS Analytics tools to improve the in-flight customer experience, in part by applying machine intelligence to user behavior (e.g., the devices they use) and flight information (flight length and destination city).

Those insights feed into Amazon Web Services, which Gogo’s data scientists use to build project dashboards that are shared with its partners. The reports have already prompted two airlines to redesign their in-flight entertainment systems and sparked others to investigate ways to personalize their systems.

“[We leverage the] data to get a better sense of what’s going on,” Wicklein said.