Meta Muse, OTAs and the future of direct hotel bookings
An agent can already compare hotels and carry a booking through to payment. If a hotel does not present an offer an agent can understand, compare and confirm, the agent will choose the path that does.
Meta launched Muse in the United States on September 8, 2026. It is a personal agent: it can research, compare options and carry out tasks with user approval, not merely answer questions. In travel, its connection to Duffel allows live flight-inventory search, comparison and booking management. Hotels are less structured. Early testing reported by Skift found Muse browsing consumer sites such as Expedia and Hotels.com.
That difference matters. Flights have a defined data path; hotels currently depend on where an agent finds enough information to complete a request. The channel that presents availability, all-in price, terms and a dependable booking route has an advantage. Today, OTAs do that at a scale no independent hotel should dismiss.
The shallow reading is that agents will remove intermediaries. The evidence does not support that claim. Expedia has announced hotel bookings through Muse for U.S. travelers and will remain the merchant of record. OTAs also have broad inventory, loyalty programs, comparable data and payment flows that already work. They are well positioned for this new distribution surface.
The hotel opportunity is not to wager on Booking or Expedia disappearing. It is to stop being a passive listing within them.
What changes when the guest does not browse
For years, the guest journey followed a familiar sequence: search, compare, read reviews, visit an OTA or hotel website, pay. An agent compresses that sequence. It can receive an instruction such as “find a hotel in Cartagena for two people, December 12–15, near the historic centre, with breakfast and flexible cancellation,” then decide which
The question shifts. “Does my website look good?” is no longer enough. A hotel needs to ask:
- Can a system determine which room is available for those dates?
- Can it identify the complete price, currency, taxes and mandatory fees?
- Can it read cancellation terms without decoding vague language?
- Can it reach the booking flow without losing dates, guests or conditions?
- Does it receive a verifiable confirmation rather than a request screen?
Those are not technical niceties. Muse’s connector guidance asks accommodation services to support availability search, complete price and terms, authorization, confirmation and follow-up for changes or cancellations. That guidance is a useful operating standard even for a hotel that does not integrate with Muse today.
The early signal: lodging does not yet have one path
Muse launched with Duffel for flights. Duffel says the connection gives access to live inventory and pricing across more than 500 airlines, with post-booking management. For hotels, no equivalent launch path was announced. Skift observed open-web browsing of consumer sites; Expedia later announced its forthcoming Muse hotel-booking integration.
For Colombia and Latin America, this does not mean Muse is already operating locally. Meta and Duffel positioned the initial launch in the U.S., and Expedia also referred to U.S. travelers. The relevant signal is that the distribution model is being tested in the market where many travel platforms establish their next integrations.
Once an agent layer sits between a guest and search, a hotel is not simply competing to be seen. It is competing to be interpretable and operable.
Direct booking is still a product decision
An owned website is not automatically a useful direct channel. A polished homepage that sends visitors to “ask us on WhatsApp” may work for a patient guest. An agent needs data and a defined next step. The same applies to a booking engine that shows a starting rate and reveals taxes, restrictions or deposits at the end.
For an independent or boutique hotel, the defensible baseline is not building its own agent. It is building a booking operation that does not depend on a person filling in the gaps.
That baseline includes:
- Current inventory and terms. Room types, maximum occupancy, relevant amenities, accessibility and restrictions cannot live in an old PDF or across disconnected channels.
- Full price before payment. The rate should separate room, taxes, mandatory fees, currency and when payment is collected.
- Visible rules. Cancellation, deposits, modifications, no-shows, minimum age, pets and check-in times should be concrete and findable.
- A short transactional path. The direct engine should preserve dates, guests and room selection through confirmation. If it needs authentication or approval, it should make the next step explicit.
- Confirmation and support. A booking reference, receipt and clear route for changes reduce uncertainty for both guest and agent acting with permission.
- Consistent data across channels. If an OTA says breakfast is included and the owned site does not, an agent does not see a branding detail. It sees a conflict.
None of this replaces OTA relationships. It makes them part of a deliberate distribution architecture rather than the only place where transactional clarity lives.
What a hotel should decide now
Start by inventorying the booking journey as it actually exists, not as it appears in a deck. Compare a real stay on the hotel website, its booking engine and two OTAs: same dates, same occupancy, same room. Record gaps in total price, terms, inventory and messaging. If the answer changes by channel, a product decision is waiting.
Next, define what belongs to the direct channel. It does not have to be a discount. It can be a more flexible policy, a clearly included benefit, a specific room, an added experience or direct pre-arrival communication. What matters is that the benefit is real, verifiable and visible before booking.
Then make exceptions legible. Premium hotels have particular rules by design: high-season minimum stays, connecting rooms, transfers, accessibility, special menus, children and pets. An agent should not make them up. The hotel needs to express them precisely and provide a handoff route when human judgment is needed.
Finally, treat integration as a commercial and operational decision. A direct connector can be valuable when there is clear demand, well-governed inventory and capacity to support changes, refunds and failures. It is not a technology badge. Before connecting a new surface, decide who controls availability, who confirms payment, what happens after a duplicate booking and who supports the guest.
The asset worth building
The advantage will not be a page “optimised for AI” packed with machine-facing language. It will be a booking asset that is clear for a person and structured for systems acting on their behalf.
Pimentone works with hotels and hospitality businesses to build owned channels that connect experience, conversion and operations. That may include transactional architecture, integrations and ongoing evolution. Technology does not replace hospitality. It keeps a valuable reservation from being lost to incomplete information or a disconnected operation.
Muse is an early signal, not a verdict on OTAs. Platforms will retain a powerful role. But a hotel that exists only as third-party inventory gives up the conversation before it begins. A hotel that can present a precise offer and carry it through to confirmation keeps an option open: the next agent can book it direct as well.
References
- 01
- 02Connector guidelines
Muse Connector Platform
- 03
- 04
- 05