Part 1: We still think hotels have the advantage
Part 1: We still think hotels have the advantage
[Edit: added a new intro and lightly modified the original post since ChatGPT released Apps for ChatGPT about a week after the ACP announcement. You can read Part 2 here.]
It’s been an interesting seven days in hospitality AI. The things we’ve been talking about for years—that someday a guest will plan and book an entire trip inside ChatGPT—no longer feel merely inevitable. They feel imminent.
At risk of subtweeting some of our tech company peers: we're not worried that hotels will lose the AI platform shift. We don't think inaction now means hotels lose distribution forever. Speed isn't everything. Because hotels actually have some pretty fundamental advantages versus the OTAs on when it comes to AI and how consumers will want to use this technology.
Now admittedly, I drafted the first half of this post right after ChatGPT released Agentic Commerce Protocol (ACP) last Monday. Since then, Apps in ChatGPT were announced just yesterday—and two of the first seven apps are from Booking and Expedia.
But have no fear. I'm here to say this is all going according to plan.
The pre-Apps take: the long tail + ACP
On Tuesday, after ChatGPT released Agentic Commerce Protocol ACP, the tech industry's best analyst and the Father of Aggregation Theory, Ben Thompson, laid out what this meant for for the future of commerce. At one point, he discusses how AI search is fundamentally different than Google search because it rewards the long-tail of commercial options. While traditional search favored Amazon, AI search favors small companies with distinctive products who distribute direct to consumer. I'll quote the whole paragraph, because he's laid this out more clearly than I can:
There are certainly some AI queries that are basically search queries, but many commerce related queries come higher up in the funnel: you have a vague idea of what you want, or know the problem you are trying to solve, and what you are looking for is help in refining that vagueness into a concrete product that you can purchase. Sometimes the answer is a popular product, but what is compelling about AI as a commerce driver is that it has the capability of finding products in the long tail that you might never have found normally but which the AI, with its vast corpus of knowledge, can surface for you. These kinds of products can certainly be found on Amazon or Walmart, thanks in particular to their large third-party merchant platforms, but the chief home for the long tail is Shopify (for companies) and Etsy (for individual sellers).
Let's substitute Booking and Expedia for Amazon and Walmart in the above quote and "hotels" for "products." So the question is, will direct-to-consumer (direct-to-guest), which has been a too-small slice of hotel distribution for the past 30 years, return in the era of AI distribution and search.
OTAs won the last era by aggregating everything, creating an infinite scroll, and buying the top of the funnel. That’s a human-eyeballs strategy. AI assistants can read far more than a traveler ever could—and they prefer authoritative sources with a clean handoff to buy. Aggregation isn’t the advantage; being the source is.
That’s why ACP matters. It gives hotels a standards-based way to surface what makes them bookable—rooms, rates, availability, perks—and to complete the purchase without inserting a new middleman. And because ACP is a protocol, the same truth can power any AI surface that matters next. Hotels aren’t merely “competitive” with OTAs here; they can be preferred when the assistant is choosing who to trust and where to send the guest.
An individual hotel is, by definition, the long tail. The long tail is where AI shines. And that’s why direct wins.
Kismet is the first-party AI storefront, catalog, and checkout for hotels. We help hotels to set up and manage their first-party presence for AI search and booking. Always your brand, never ours.
If you’re ready to sell through AI, we’re ready to help.