Why Travel's AI Assistants Look Seller-Side

Mark Zuckerberg’s new manifesto, “The Future is for Everyone,” sets out a simple principle: an AI agent should work for the person using it, not for the company that built it. In travel, that principle is largely absent. The latest industry commentary argues that the AI assistants already deployed by travel companies are designed first to sell, not to serve.

Examples are specific: Booking.com’s trip planner, Expedia’s Romie, airline chatbots and hotel concierge tools are built around direct bookings, upsells and loyalty. They are seller-side agents, because their economic purpose is to move inventory and capture demand for the brand, not to find the best outcome for the traveler.

The implications are substantial if a genuinely traveler-side agent wins trust. Urgency banners, retargeting and email campaigns would lose power. Suppliers would instead be rewarded for transparent pricing, agent-accessible inventory and strong product information. But the analysis flags an unresolved conflict: Meta earned roughly $196 billion in 2025, almost entirely from advertising, and the history of OTAs, metasearch engines and Google shows that consumer-friendly layers eventually monetize by selling access to demand.

The open question is therefore straightforward: can any personal agent remain loyal to the traveler once it has to become a business?

Advertisement

A Traveler-Side Agent Would Rewrite Distribution

Booking.com, Expedia and the Seller-Side Default

The pattern is structural, not accidental. Booking.com’s planner, Expedia’s Romie and similar airline or hotel chatbots are tied to a commercial loop: recommend, convert and retain the customer inside the brand’s own ecosystem. That is not deception by itself—many are transparent booking tools—but it means the traveler is the product being steered, not the client being served. The distinction matters because the current tools optimize for the seller’s conversion rate, not for the best available option across all suppliers.

A Traveler-Side Agent Would Reverse the Persuasion Economics

A trusted agent aligned with the traveler would change the economics of persuasion. Tactics such as urgency messaging, retargeting and broad email campaigns rely on capturing attention and creating pressure. An agent that compares options dispassionately would strip much of that value. Suppliers with clear cancellation terms, accurate pricing, machine-readable inventory and detailed product descriptions would gain, while suppliers that rely mainly on marketing spend could lose visibility. This is the analytical core: distribution would reward product quality and data availability over demand-capture technology.

Meta’s Advertising Model Is the Unresolved Tension

Meta’s scale makes the conflict acute. With roughly $196 billion in revenue in 2025 and an advertising-led model, any Meta-built agent would face the same pattern seen with OTAs, metasearch and Google: a consumer-friendly layer eventually sells position or access to demand. The article stops short of predicting that outcome, but the historical parallel is the reason a “personal” agent cannot simply be assumed to remain loyal. The business model, not the marketing language, will determine whose interests the agent ultimately serves.

What Travel Suppliers and AI Builders Should Do Now

The argument is a strategic warning rather than a product launch, so the practical steps fall to suppliers, AI builders and investors rather than to travelers.

  • Travel suppliers should measure how much direct demand currently depends on urgency banners, retargeting and email campaigns; these persuasion layers are the first to weaken if a trusted traveler-side agent scales.
  • Booking.com and Expedia should disclose whether their AI assistants can rank recommendations based on supplier payments or margin, because the OTA-metasearch-Google history described in the piece makes undisclosed placement the main trust risk.
  • Product and revenue teams should treat agent-readable inventory and clear product information as commercial assets; the analysis identifies those as the factors that would favor suppliers in a traveler-side distribution model.
  • Investors and boards evaluating travel AI tools should ask who pays when a recommendation is made; if the answer is seller commission or upsell margin, the tool is seller-side under the definition used here.
  • Because Meta’s $196 billion advertising base creates a structural conflict, any Meta travel agent should be evaluated on whether recommendation access is or can be monetized, not on its stated loyalty to the user.

Risk & Opportunity Assessment

Commercial RiskMediumIf a traveler-side agent wins trust, suppliers that depend on urgency banners, retargeting and upsell conversion could see demand-capture value fall; no immediate loss is demonstrated.
Competitive RiskHighBooking.com, Expedia and other seller-side AI tools would face a new intermediary that rewards transparent pricing and accessible inventory, potentially shifting share away from large marketing budgets.
Regulatory RiskLowThe article raises no specific regulatory action; risk would only emerge if AI recommendation ranking attracts disclosure or consumer-protection rules.
Reputation RiskMediumMeta and other ad-funded platforms face the historical OTA-metasearch-Google pattern, making claims of a traveler-loyal agent vulnerable to skepticism if recommendation access is later sold.
Technology DisruptionHighA genuinely traveler-aligned AI agent would be a new distribution layer, displacing persuasion-based conversion and changing how suppliers compete.
Commercial OpportunityHighSuppliers with clear pricing, agent-accessible inventory and strong product information could gain visibility and reduce reliance on paid demand capture if such an agent scales.