United Airlines Accelerates Digital Transformation with AWS Bedrock to Power Enterprise-Scale Generative AI

Key Metrics

• Successfully implemented AWS Bedrock in under 9 months as an early preview customer
• Planning to add over 700 aircraft (nearly doubling fleet size) with AI-enhanced operations
• Reduced development time for AI applications through reusable templates and accelerators
• Eliminated need for 180-degree pivot of data and ML engineering teams by leveraging existing AWS MLOps platform

Background

United Airlines is one of the world's largest airlines with a mission to be the best in aviation—not just in size or route network, but in customer experience, customer loyalty, and employee engagement. As part of their aggressive growth strategy, United is planning to add approximately 700 aircraft to their fleet over the next few years, while maintaining their "Good Leads the Way" motto that emphasizes community service and customer-centricity.

The Challenge

When generative AI exploded onto the scene in early 2023, United Airlines found themselves reacting to a technology revolution rather than driving it. They faced significant pressure as vendors and partners began approaching business users across the enterprise with solutions promising to solve all their problems through generative AI.

I think it would be a safe bet to say that when 2023 dawned that most of our companies here did not have a plan for generative AI. It was more that generative AI hit us than we had a plan for it. But we had to react quickly. — Sanjay Nair, Managing Director - Data Engineering at United Airlines

While recognizing the vast opportunities presented by generative AI, United also understood the potential risks and pitfalls of this emerging technology. As a complex enterprise with numerous operational domains—network operations, flight operations, revenue management, technical operations, and maintenance—they needed a structured approach that would enable responsible innovation while preventing security nightmares and uncoordinated point solutions across the organization.

The airline needed a solution that would allow them to move quickly, integrate easily with existing systems, and provide the flexibility to experiment with different models while maintaining centralized governance and security controls.

The Solution

United Airlines selected AWS Bedrock as their generative AI platform after careful evaluation with strategic partners. As one of the first companies to preview Bedrock, United was impressed with several key features that aligned perfectly with their requirements.

AWS and Bedrock provides the foundational infrastructure and they'll continue to innovate and support that. And what we believe we'll focus on is the data to create the data knowledge base and then focus on the prompts and the responses to continue to generate significant value for the business. — Sanjay Nair, Managing Director - Data Engineering at United Airlines

Three critical factors drove United's decision to implement AWS Bedrock:

First, as a fully managed AWS service, Bedrock integrated seamlessly with United's existing cloud infrastructure, eliminating the need for extensive rearchitecting. Second, Bedrock's API invocation framework made it accessible to business users and application development partners, accelerating adoption. Third, the platform offered choice among multiple foundation models, allowing United to experiment and select the optimal model for each specific use case.

Additionally, Bedrock addressed United's security concerns by providing built-in responsible AI features, centralized operations capabilities, and data privacy guarantees that their information would not be used to train foundation models.

Strategy

United implemented AWS Bedrock as part of a comprehensive enterprise generative AI strategy that balanced innovation with governance. Their approach leveraged their existing data infrastructure while adding new capabilities specifically designed for generative AI applications.

Building on Existing MLOps Foundation

Rather than starting from scratch, United integrated Bedrock with their existing United Data Hub and MLOps platform that they had modernized on AWS over the previous 15 months. This strategy allowed them to accelerate implementation by extending established platforms rather than building entirely new systems.

The team created a technical architecture that connected AWS Bedrock to their data platform, enabling API invocation, model selection, and extension into specialized features like knowledge bases and vector databases. This foundation supported document embedding, context preservation, and conversational memory for their LLM applications.

Creating Reusable Templates and Accelerators

To scale development across the enterprise, United developed templates and accelerators for common use cases like summarization and chatbot development. These templates significantly reduced custom work required from business users and application development teams, while accelerating implementation times for new applications.

We added some accelerators and adapters to this framework, which you see as templates here. These templates are really helping our use cases and our business users. And this will continue to pay forward as more and more use cases come. — Sanjay Nair, Managing Director - Data Engineering at United Airlines

The team also implemented service logging to capture prompts and responses, storing this data in S3 within their United Data Hub. This approach allowed them to continuously improve and fine-tune their models based on actual usage patterns.

Establishing a Use Case Prioritization Framework

United created a structured program to intake and evaluate generative AI use cases, prioritizing them based on a matrix of value and complexity. They initially focused on high-value, low-complexity implementations to build experience before tackling more complex scenarios.

Their customer-centric strategy targeted four primary use case categories:

1. Summarization and rewriting: Simplifying complex policies and business rules to provide relevant information to employees and customers
2. Conversational interfaces: Enhancing one-on-one interactions through their award-winning mobile app and agent tools
3. Sentiment analysis: Converting vast customer feedback into actionable insights to improve satisfaction and experience
4. Multimodal LLM applications: Combining text, speech, and other modalities to create more immersive customer and employee experiences

Results

United Airlines has successfully established a robust enterprise foundation for generative AI that balances innovation with governance. By partnering with AWS and implementing Bedrock, they've created a platform that will support hundreds of use cases while maintaining security, responsible AI practices, and cost efficiency.

The implementation has already delivered significant strategic advantages. United did not need to completely retrain or reorganize their data and ML engineering teams, as the Bedrock integration built upon their existing AWS MLOps platform. This approach accelerated delivery timelines and reduced disruption to ongoing operations.

Looking forward, United is focused on three key areas for continued advancement. First, they're strengthening responsible AI and security with guardrails, prompt engineering, and retrieval-augmented generation to enhance foundation models with proprietary knowledge. Second, they're optimizing their platform through hardware and database optimizations to manage costs as use cases scale. Finally, they're exploring multimodal models and fine-tuning approaches to create United-specific models for targeted use cases.

Most importantly, United has created a partnership model where AWS continues to innovate on the Bedrock platform while United's data engineers and ML specialists partner with business units to deliver improved customer experiences and operational efficiencies.

Our focus is going to be on our use cases and our data while we partner with AWS on enhancing and working towards a more innovative Bedrock platform so that our data engineers and our ML engineers can focus on partnering with our businesses to create business efficiency and more importantly, create the customer experiences that enable our customers to keep coming back and flying on United Airlines. — Sanjay Nair, Managing Director - Data Engineering at United Airlines