The travel industry is facing a fundamental identity crisis driven by artificial intelligence. As Online Travel Agencies (OTAs) and tech giants race to integrate AI, a strategic divide is forming. The central question is no longer how companies will use AI, but rather what their fundamental role will be in a world where machines handle the planning.
Industry players are currently making bets across five distinct layers of the “AI Travel Stack.” Each layer represents a different vision of who will ultimately control the traveler’s journey.
1. The Model Layer: Building the Brains
The Model Layer focuses on the raw intelligence behind the interface. Instead of relying on generic models like GPT-4, companies are attempting to build proprietary Large Language Models (LLMs) specifically tuned for the travel industry.
A recent job posting from Booking Holdings reveals this strategy in action. The company is seeking a machine learning manager in Amsterdam to lead a team dedicated to building generative AI foundation models trained on Booking’s own massive datasets.
Why this matters: Generic AI models often struggle with the specific nuances of travel—such as real-time availability, complex routing, or local hospitality contexts. By owning the model, a company ensures its “brain” is uniquely optimized for travel, creating a competitive moat of specialized knowledge.
2. The Orchestration Layer: The Master Coordinator
The Orchestration Layer sits between the raw model and the user. These companies don’t necessarily build the LLM, but they build the logic that tells the model how to act. They manage the “reasoning” process—deciding when to call a flight API, when to check a hotel database, and how to synthesize that information into a coherent plan.
3. The Product Layer: The User Experience
The Product Layer is where most travelers currently interact with technology. This layer focuses on the interface—the apps, websites, and chatbots. The goal here is to wrap AI capabilities in a seamless, user-friendly design. Companies in this layer win by providing the most intuitive and frictionless experience, regardless of which model is running in the background.
4. The Legibility Layer: The Data Foundation
Running beneath all these layers is the Legibility Layer, managed by infrastructure players. This layer is responsible for making travel data—which is often fragmented, messy, and siloed—readable and usable for AI. Without high-quality
