JetBlue is actively using artificial intelligence and machine learning across its business and actively using generative AI for its internal operations and ultimately revenue-producing products.

Speaking at the Databricks Data + AI Summit, Sai Ravuru, Senior Manager of Data Science and Analytics at the airline, walked through how the company was using Databricks Lakehouse on multiple fronts. Databricks launched a series of new additions to its platform and said it will acquire MosaicML

"Over the last two years, we've made investments in data science and data refinement so raw data is continuously hydrated and reliable," said Ravuru. He said that AI and machine learning teams at JetBlue work alongside data scientists. "AI/ML scouts for the next use case before handing off to the data science team," explained Ravuru, who noted the goal for JetBlue is to be the most data-driven airline.

Ravuru said that data touches every part of JetBlue's business including operations, commercial and support functions. JetBlue is creating a unified digital twin of its business with cross-team collaboration and process-driven data science fueled by data from multiple systems.

Databricks Lakehouse is what absorbs and creates modeling across JetBlue's data footprint.


The airline has leveraged Databricks' platform to create an ecosystem of models called BlueSky to enable decision making. "The BlueSky product was built from scratch internally," said Ravuru. "It is a continually refreshed network with embedded LLM and real-time components for frontline staff."

BlueSky serves as JetBlue's AI-driven operating system.

Ravuru also said that JetBlue has created a unified LLM called BlueBot that uses open-source models complemented by corporate data integrated with BlueSky. BlueBot can be used by all teams at JetBlue since access to data is governed by role. For instance, the finance team may see data from SAP and regulatory filings, a new employee may just be served FAQs, and operations would see maintenance information, explained Ravuru.

"BlueBot brings crew members much closer to data and insights without change management," he said.

JetBlue is using Databricks for generative AI use cases that are experimental as well as production.

What's next? JetBlue is looking at LLMs to create new revenue channels so customers "can book from BlueBot or plan trips better." In addition, JetBlue is looking at efficiency gains by using LLMs to provide the "technical operations team with WebMD style diagnoses for each and every aircraft."