The OTAs Aren't Being Disintermediated by AI. They're Becoming the Infrastructure Under It.
Source: Rajeev Goswami, CEO of WWStay; WWStay subscriber brief (Substack) · 2026-08-31
A prospect asked me last week, in a discovery call for a project-housing deal, whether AI agents were going to make our OTA relationships irrelevant within two years. I gave the answer everyone in the industry gives right now: probably, eventually. Then I went and actually checked the 2026 data, and I don’t think that answer is right anymore.
Independent hotels’ RevPAR fell 5.4% in 2025 while their OTA dependence climbed to 63.4% of bookings — worse, not better, in the year AI was supposed to start fixing this. OpenAI’s quiet retreat from ChatGPT checkout in March turned out to be the tell: even when an AI agent books through Booking.com or Expedia’s API instead of their website, those platforms stay merchant of record and keep the commission. AI is removing the storefront, not the infrastructure underneath it.
I started digging into Expedia’s B2B numbers expecting a footnote and found the real story: their B2B segment — the part powering other companies’ booking tools, including TMC (travel management company) platforms your program may already sit inside — grew 24% year-over-year last quarter and is now 38% of total revenue. Branded chains’ OTA channel share has barely budged in a decade of “book direct” campaigns.
Nobody’s actually measuring whether an AI layer on a TMC platform changes the underlying OTA-supply relationship. The dashboards travel managers watch track booking behavior, not who’s still collecting the commission underneath it. I started asking which of my own client relationships run through exactly this kind of embedded OTA supply — and then I started worrying I already knew the answer.
If your OTA content sits inside your TMC’s platform today, is AI actually reducing that dependency — or just making it harder to see?
Lines of inquiry this paper opens 24
Research framings built by reading the notes related to this paper — the questions it feeds into.
Do AI shopping agents exhibit systematic bias in recommendations?- How does caching refresh cadence affect fare accuracy for consumers?
- Which retail categories are most vulnerable to agent-driven basket fragmentation?
- Does infrastructure control over fares shift market power away from airlines?
- How much ad revenue does Amazon lose when agents skip sponsored listings?
- Can retailers use off-site agents to compete without losing customer loyalty?
- What happens to traditional retailer-shopper relationships as third-party agents grow?
- How might card networks recognize agent-operator liability to shift risk away from merchants?
- Which actors currently control the largest shopping agent platforms?
- What makes agent-to-agent negotiation fundamentally different from agent-assisted shopping?
- Can AI layers on TMC platforms change underlying OTA supply relationships?
- How does travel commerce become the first proving ground for agents?
- How do booking aggregators like Booking.com justify their 14-15 percent take rate?
- Why would merchants accept higher agent take rates as they mature?
- Can centralized data ownership alone move travel agents out of pilot mode?
- How does MCP's zero-commission model compare to current OTA commission structures?
- Where exactly does fulfillment infrastructure value concentrate in travel booking?
- Why do different OTA categories show uneven performance against AI disruption?