The Five Layers of the AI Travel Stack, and the Fight Over Which One Matters

Why It Matters

Control of a dominant AI layer could redefine OTA value propositions and shift the balance of power toward platform owners, reshaping how travelers discover and book trips.

Key Takeaways

• Booking builds travel‑specific LLMs to own model layer
• Expedia invests in orchestration to combine multiple AI models
• Airbnb focuses on product layer, enhancing AI‑driven user experience
• Sabre develops legibility layer, making travel data machine‑readable
• Platform giants could render OTA layers irrelevant via AI OS

Pulse Analysis

The rise of generative AI has forced online travel agencies to rethink their technology stack, breaking it down into five distinct layers. At the base, the model layer houses large language models trained on proprietary data, promising highly tailored recommendations. Above that, the orchestration layer stitches together multiple models, enabling dynamic pricing and inventory management. The product layer translates AI insights into consumer‑facing features, while the legibility layer ensures data is structured and machine‑readable for downstream applications. Overarching all is an OS/platform layer that could dictate the ultimate user interface, potentially sidelining traditional OTA offerings.

Each OTA’s strategic focus reflects its core competencies and risk appetite. Booking Holdings and Trip.com are pouring resources into custom travel‑specific LLMs, betting that owning the model will lock in data advantages and create defensible IP. Expedia, by contrast, leans on orchestration, positioning itself as a broker that can integrate best‑of‑breed models from partners, reducing development risk. Airbnb’s emphasis on the product layer aims to embed AI directly into the guest experience, leveraging its brand to differentiate. Meanwhile, infrastructure stalwarts like Sabre invest in the legibility layer, standardizing travel data to become the lingua franca for AI services. These divergent bets illustrate a broader industry dilemma: whether to own the engine, the conductor, or the performance.

The ultimate winner may be the platform layer, where tech giants such as Apple, Google, and Microsoft could embed AI travel assistants directly into their ecosystems. If travelers begin to rely on a single AI‑driven interface for itinerary planning, booking, and post‑trip services, the relevance of OTA‑specific layers could diminish. This scenario would shift revenue streams from commission‑based models to platform‑centric monetization, compelling OTAs to either integrate with these ecosystems or reinvent their value proposition. Stakeholders must monitor how AI governance, data privacy, and consumer trust evolve, as these factors will determine whether the AI travel stack reshapes the market or merely adds another layer of complexity.