More Channels Won't Get You Booked by AI

Someone running a handful of short-term rentals through Guesty typed a question into the chat on this site last week. She wants the software to connect to more agentic channels. It's a reasonable thing to want. It's also the wrong fix, and working out why tells you most of what you need to know about where this is heading.

For fifteen years, "connecting to more channels" has meant the same thing for a small operator. You plug your calendar into Airbnb, Booking.com, Vrbo, Expedia, maybe a regional player, and a piece of software makes sure the same room doesn't get sold twice on a Saturday night. That's channel management. It's plumbing. It solved a real problem: double bookings, stale availability, a guest showing up to a room that was already gone. Nobody's nostalgic for the spreadsheet era it replaced.

Now the same instinct is being pointed at a different problem, and it doesn't fit. "Connect me to more agentic channels" sounds like an upgrade to that old plumbing. It's actually a request to be piped into someone else's AI travel agent, on their terms, inside their inventory, ranked by their logic. That's not the same job at all.

What does "agentic channel" even mean right now?

Look at what's actually being built. Trip.com and Mastercard are expanding TripGenie from a search tool into something that can compare, select, and complete a booking with payment, inside Trip.com's own environment. Expedia scrapped its attempt at one master chatbot and is now building a cluster of specialized agents, backed by its acquisition of the trip-planning startup Layla, all still living inside Expedia's own funnel. Trip.Biz built a four-agent suite for corporate travel, and one of its stated goals is keeping bookings inside the company's own booking tool instead of leaking out to a direct site. Travelxp's Marco can discover, plan, book, and pay for a trip on a traveler's behalf, but it does it by assembling that trip from Travelxp's own commercial relationships.

Every one of these is being described in the press as "agentic booking." Every one of them is also a closed loop. The agent doesn't go looking across the open web for the best halibut dinner in Sonoma or the sharpest fly-fishing guide on the Yampa. It shops the shelf its own company stocked. Getting your listing plugged into that shelf isn't the same as being visible to an AI agent. It's being inventory inside somebody else's storefront, which happens to now have a very persuasive salesperson standing at the door.

That distinction matters more than it sounds like it should, because of what happened the last time small operators chased more channels without asking who those channels actually served.

More channels already did this once. Read the data.

A HOTREC-linked study on the European hotel market found that OTAs went from 19.7% of bookings in 2013 to 29.9% in 2026. In the same window, direct bookings dropped from 57.6% to 51.3%. And within that OTA share, Booking Holdings alone grew from 60% of OTA bookings to nearly 70%. Booking and Expedia together now account for more than 85% of OTA bookings in Europe. That's more than a decade of hotels connecting to more channels, in good faith, trying to reach more guests, and watching the two biggest gatekeepers tighten their grip the entire time.

Now bolt an AI layer on top of that same structure and ask yourself what changes. TripGenie doesn't dilute Trip.com's power. It automates it. Expedia's multi-agent architecture doesn't open Expedia up. It makes the closed loop faster and harder to see inside. If plugging into more channels concentrated power in a handful of platforms before any of this had a chatbot attached to it, plugging into more "agentic channels" will do the same thing with better manners and a friendlier voice.

There's a reason independent hotels and guesthouses read that 85% number and feel their stomach drop a little. It's not abstract. It's the difference between a guest who found you and a guest who got assigned to you by an algorithm optimizing for something you'll never see.

Who is the agent actually working for?

Here's the question worth sitting with, and it's not the one about how many channels you're wired into. It's who the agent is working for when it books you.

An agent built by Booking, Expedia, or Trip.com works for the platform first. It might be genuinely useful to the traveler. It might even be polite about it. But its job, structurally, is to keep the transaction inside the walls that pay for its development. A guest's own assistant, the general-purpose kind that isn't owned by an OTA, has a different job: find the actual best answer to "private chef in Charleston, six people, Friday, budget four hundred," wherever that answer lives. Those two agents can produce very different shortlists for the same request, and only one of them has any reason to consider you if you're not already sitting on its shelf.

The SkyMatch model for helicopter charters is a small, useful example of where the industry is drawing the line right now. Their AI agents can research routes and assemble a request, but only a licensed human operator can approve and finalize the booking. It's a safety rule for a genuinely risky activity, but it also exposes something true about every version of this: somewhere in the chain, a human confirms the thing that actually matters. For a hotel room, that confirmation increasingly happens inside a platform's own AI. For your private dinner, your dawn boat trip, your two-night cooking retreat, it still has to happen somewhere. The only question is whether that somewhere is your own site, legible to any agent that comes looking, or a platform's showroom that decides which agents get to see you at all.

That's the real fork. Not "how many agentic channels am I connected to," but "can an agent that isn't owned by anyone find me without going through a platform's toll booth first."

What actually helps here

The fix isn't glamorous, and it isn't a new integration to buy. It starts with making your own site and your own listing legible in plain, structured terms: what you offer, what it costs, when it's available, what the cancellation terms actually are. Not a PDF, not a paragraph of atmosphere copy, not a calendar widget that only a human with a mouse can operate. Facts, stated plainly, in a form software can read without guessing.

Then stop treating every OTA-owned bot as a customer acquisition strategy and start treating it as what it is: a shelf you don't control, in a store that takes a cut and decides your position in it. Being listed there is fine. Depending on it entirely, especially once that shelf has an AI clerk deciding who gets recommended, is how a chef ends up cooking eight covers for a platform's margin instead of five for a guest who actually chose him.

And if you want to know, concretely, what an independent agent sees when it goes looking for you outside anyone's walled garden, that's worth checking directly rather than guessing. It's the one thing bookaible is built to show you plainly, no funnel attached, no platform to switch to.

The guest who typed "private chef, six people, Friday" into an assistant that isn't owned by Expedia or Trip.com is going to get an answer from somewhere. The operator who spent a decade optimizing for more channels learned that being everywhere on someone else's terms is not the same as being findable on your own. Direct bookings didn't fall because guests stopped wanting a direct relationship. They fell because the toll booths got better at looking like the whole road. An AI travel agent that works for the platform is just a faster version of the same booth, and no amount of extra plumbing into it is going to change who it was built to serve.