United Airlines' transition to cloud computing, particularly through its partnership with Amazon Web Services (AWS), has been pivotal in enabling its generative AI capabilities.

As one of the earliest adopters of the Amazon SageMaker ML platform, United is leveraging its own United Data Hub and the Mars ML platform to develop innovative generative AI models.

The airline is utilizing the AWS Generative AI Innovation Center to enhance its digital operations, with the potential to automate many customer communication tasks.

CIO Jason Birnbaum has rolled out several AI models, including United ChatGPT for secure internal use and the customer-facing LLM 'Every Flight Has a Story,' which has improved communication about flight delays and boosted customer satisfaction by 6%.

Launched earlier in 2024, the 'Every Flight Has a Story' service provides customers with timely and detailed updates on flight delays, enhancing transparency and reducing confusion.

This LLM aims to provide specific information about flight disruptions, such as exact weather conditions causing delays, rather than vague announcements, which helps alleviate customer anxiety.

United Airlines has already sent over 100,000 digital notifications using this LLM and has identified approximately 90 additional use cases for AI across its operations.

As of December 2024, the adoption of generative AI in production at United has increased significantly from 4% in March 2023 to 21%, reflecting a broader trend among enterprises to advance beyond pilot phases.

The implementation of SageMaker and Bedrock is expected to greatly enhance United's data product development, fostering a data-driven culture within the organization.

An upcoming beta application will summarize essential operational data for staff during shift changes, aiming to improve efficiency during handovers.

In addition to the flight-status LLM, United has developed LLMs to streamline procurement processes and enhance communication between managers and employees.

Overall, United Airlines is advancing its use of generative AI with production-level large language models built on a robust cloud-based data infrastructure.