AI's new gatekeepers: How Booking.com and Expedia are hijacking the future of travel
Here's the uncomfortable truth: a new duopoly is forming in plain sight. Booking.com and Expedia moved first into AI assistants, and that early move could turn them into the default middlemen of the next decade.
If that sticks, hotels will pay to access their own guests-again. Different interface, same toll booth.
What's actually happening
AI assistants aren't just answering questions. They decide what's recommended, who shows up, and who gets booked-often without sending a traveler to your site.
OpenAI's new ecosystem gives big OTAs a privileged lane. Perplexity compounds it. And if preferred placements become the norm, distribution power concentrates fast. This isn't a neutral interface; it's a programmable funnel controlled by a few players.
Why the DMA matters-and where it misses
The EU's Digital Markets Act (DMA) forces big platforms to "own the road" without tilting the traffic. It protects fair access, clearer ranking, and limits on self-preferencing. That's good policy.
The gap: standalone AI assistants aren't fully covered. Gemini, ChatGPT, and others can act like distribution gateways without DMA-grade guardrails. If that continues, smaller brands get buried in opaque rankings, and paid placements silently become the default route to demand.
The stakes, by the numbers
- Expedia Group spans 200+ sites, 75+ markets, 40+ languages, with ~10M daily visitors and 1B+ annual bookings.
- Put that scale inside AI assistants with 100M daily active users and you're looking at ~9.6B signal events per day feeding the rankings.
- This week, ~800M people will use major assistants. In five years, AI agents could exist across devices, cars, TVs, and apps-tens of billions of touchpoints.
Give a few companies preferential access to that data and interface, and independent hotels become permanent renters.
What policymakers must do
- 1) Extend DMA protections to standalone AI assistants acting as distribution gateways.
- 2) Mandate non-discriminatory access for qualified suppliers based on published technical criteria.
- 3) Require transparent ranking logic and clear labels for paid placements.
- 4) Limit cross-context data use for sensitive or identifying signals.
- 5) Enable easy switching and multi-homing for business users.
What the industry should do next
- 6) Make ARI portable: publish availability, rates, and inventory in machine-readable formats for assistants to access direct offers.
- 7) Publish loyalty and offer rules as structured policies instead of hidden PDFs.
- 8) Instrument your site for agents: one site for people, one for code (schemas, APIs, MCP adapters).
- 9) Demand clear "sponsored" labels inside any AI interface (no gray areas).
- 10) Support DMA-style rules for standalone assistants-and say so publicly.
Practical playbook for hoteliers (start this quarter)
- Standardize your data: adopt consistent schemas for rooms, rates, policies, fees, and inclusions. Remove ambiguity that hurts rankings.
- Ship a clean, assistant-ready ARI feed. Minimize blackout logic and exceptions that confuse agents.
- Expose deals and loyalty rules as structured policies: stay length, promo eligibility, blackout dates, and stackability.
- Create an agent-safe booking path with lightweight APIs and clear state handling. Cut steps. Cut friction.
- Instrument every session: track referral source, assistant ID, prompt class, and outcome. Feed wins back into your ranking signals.
- Balance parity with direct value: extras, flexible terms, or credits that assistants can parse-not banners only humans see.
- Protect your first-party data: define what can be shared, with whom, and for how long. No blanket access.
- Test assistant placements like media buys: measure ROAS, CAC, and cannibalization before scaling.
If we don't act
Whoever controls the interface controls the guest. We learned that already. The AI layer just makes it faster, quieter, and harder to spot.
This isn't about resisting progress. It's about fair lanes, clear labels, and direct access for brands that do the work.
Want your team ready for the assistant era?
Upskill your marketing and eCommerce teams on AI assistants, prompts that convert, and agent-friendly data structures. Start with focused training built for busy operators.
About Agentic Hospitality
Agentic Hospitality is an infrastructure-level AI Cloud platform built to help hotels and resorts win direct bookings through AI-native channels and drive profitable revenue. Developed with Brewer Digital and deployed on Google Cloud and Vertex AI, it turns guest interactions into intent signals and real-time orchestration that grows loyalty.
The platform draws on 12 years of commerce architecture across leading hospitality brands and scales from independents to enterprise. It's powered by a best-in-class partner ecosystem spanning AI, automation, data orchestration, and hospitality infrastructure.
Players include:
- Google Cloud (Vertex AI, Model Garden, identity, OpenAI API compatibility)
- Brewer Digital Marketing (Schema Adapter, TravelOS Model Context Protocol, Booking Engine Adapter)
- Amadeus (CRS)
- Okta, Auth0 (SSO)
- Salesforce, Snowflake (CRM)
- Cloudbeds, Infor, Oracle Hospitality, Maestro (PMS)
- PayPal, Braintree (Payments)
- Quantum Metric (Session replay, Felix AI summaries, behavioral analytics)
- Ad agencies (Dynamic CRM-based personalization and paid media optimization)
- Little Buddy Agency (Human-AI hybrid creative and automation)
The window is closing. Get your data in order, your stack instrumented, and your voice heard in policy discussions-before the gate closes for good.
Your membership also unlocks:
