25th of September 2026: Here are a few of the stories that caught our attention over the week, with some of our takeaways.
Story 1
Expedia Plugs Into Meta’s Muse – and Makes Itself the Agent’s Supplier
PhocusWire reports that Expedia will let US travellers search for and book hotels through Muse, the personal AI agent Meta launched earlier this month. A traveller describes dates, destination and preferences, Muse returns options and the booking completes inside the agent – with Expedia remaining merchant of record and payment handled in Muse. No launch date has been given.
On 23 September, as 24/7 Wall St. reported, Expedia shares fell around 7%, Airbnb 6% and Booking Holdings 5% on fears that agents like Muse could bypass aggregators altogether. Yet Expedia hasn’t been bypassed; it has volunteered to become the supply layer underneath the agent – the same role it took in Google’s AI Mode hotel booking launch in August, alongside Booking.com, Marriott and others.
That distinction matters for hotels. When I was at Starwood, the fight was over who owned the guest at the point of search. In an agent-led journey the question becomes who owns the guest at the point of intent – and here the answer is Meta, with Expedia holding the transaction and the hotel receiving a reservation stripped of the conversation that produced it. Expedia’s Xavier Amatriain calls the future a “composed, multi-agent system”. Don’t you think hotels need a seat in that composition, rather than a line in someone else’s inventory feed?
The practical point is that the first agents are choosing large, well-connected supply partners. Hotels that want to be bookable directly in these environments need clean content, live rates and a connectivity route an agent can actually use – otherwise they default to being sold through someone else’s pipe, at someone else’s commission. No need for alarm yet – the panic about AI and hotel bookings still runs ahead of the data. But the window to prepare won’t stay open indefinitely.
Story 2
HBX Takes Its Virtuoso Partnership Global – and Bets on the Luxury Advisor
Travolution reports that HBX Group has expanded its partnership with Virtuoso from its North American base, which dates back to 2001, to a global agreement, giving Virtuoso member agencies worldwide access to Bedsonline’s full portfolio. That means premium transfers, car rental, experiences and activities on top of accommodation, plus Bedsonline’s Luxury Collection of high-end hotels, with The Luxurist’s concierge service.
On the surface this is a B2B partnership release. Underneath, it tells us where a bedbank thinks its future lies. HBX’s chief distribution officer David Amsellem frames the deal as a way to connect hotel partners with some of the world’s most discerning travellers – high-value, advisor-led demand, and precisely the segment least exposed to the AI agents that rattled online travel share prices this week. If agents collapse the comparison layer, a wholesaler that simply resells rooms looks vulnerable. A wholesaler embedded in the workflow of a trusted luxury advisor, packaging the hotel with the transfer and the experience, is far harder to replace.
For hotels, the question is familiar. Luxury and upper-upscale properties have always been wary of wholesale, and for good reason: net rates have a habit of leaking into public channels. Virtuoso business, by contrast, usually runs through preferred-partner agreements negotiated directly with the hotel, with amenities attached. So does a luxury hotel now want its inventory reaching Virtuoso advisors through a bedbank as well? Perhaps – but only with firm rate fences, clear visibility of which contract a booking came through and a view on whether the guest receives the same amenities either way. I once asked whether wholesalers were on a path to extinction; HBX’s answer is to move upmarket and closer to the advisor. Don’t you think hotels should be managing that channel as deliberately as HBX is building it?
Story 3
SiteMinder Goes After the Plumbing Beneath Hotel Pricing
Hotel News Resource carries SiteMinder’s launch of its Dynamic Commerce Engine, an AI and machine-learning layer that promises to find – and, with the hotelier’s approval, execute – revenue opportunities across both pricing and distribution. The engine will score each property using almost four billion availability, rate and inventory signals from SiteMinder’s platform, which serves 56,000 properties and powers 140 million reservations a year. It goes live over the coming months.
The headline says AI, but the real substance is plumbing. SiteMinder is explicitly targeting broken channel connections, unmapped room types and inventory sync failures – the unglamorous errors that sit underneath every rate recommendation. CEO Sankar Narayan’s point is blunt: pricing can’t be the only focus when the distribution layer underneath it is failing. I have sat through enough portfolio reviews, at Wyndham and in consulting since, to know he is right. A perfectly optimised BAR is worthless on a room type that isn’t mapped to a key OTA, and those errors can sit undetected for months.
SiteMinder’s own traveller research adds a demand-side reason to care: 51.4% of more than 12,000 respondents across 14 countries said they had abandoned an online hotel booking after a poor experience, and 26.5% blamed reviews, photos or details that didn’t feel trustworthy. As AI agents begin reading hotel data directly, inconsistency stops being cosmetic and starts being disqualifying. The broader lesson? Revenue management is steadily becoming commercial strategy, and distribution hygiene now belongs in the revenue team’s remit, not IT’s. The open question is execution – approving recommended changes still needs a human with time to look.
Story 4
RateGain’s Bhanu Chopra – When Agents Talk to Agents, Visibility Is the New Shelf Space
In an ITB Berlin interview republished by Hotel News Resource, RateGain founder Bhanu Chopra argues that AI agents have moved from support tools to active players in the sales chain, and that the first part of distribution to feel it is the aggregation and comparison layer in the middle. Six months after ITB, he says the industry is now running agents in live environments, but data readiness and organisational silos are still holding most companies back.
Two points stand out. The first is his picture of machine-to-machine distribution: a traveller’s agent querying a hotel’s pricing agent directly, and the hotel responding in real time with an offer shaped for that exact context. That is a very different model from today’s chain of connectors, mappings and delayed updates. Chopra is candid that nobody has yet settled who is accountable when two agents transact across company boundaries and something goes wrong. The second is his warning that “visibility inside that agent becomes the new shelf space”. Every technology wave in travel has promised to hand power back to suppliers, and most have produced a new intermediary instead.
Read alongside this week’s Expedia and Muse news, that warning lands. His practical advice is unglamorous and right: fix the data before buying another AI tool, make rates and availability discoverable by a machine in real time, and take down the walls between revenue, marketing and distribution. Our own hotel AI search readiness checklist makes the same case, role by role. RateGain, of course, sells the data and distribution tools that agent-readiness requires, so read the diagnosis with that in mind. It is still hard to argue with. An agent cannot act on a rate sheet that someone updates once a week. How much of your commercial logic still lives in one?
Story 5
Accor’s ALL Concierge Turns Direct Booking Into a Conversation
Hotel Dive reports that Accor has rolled out ALL Concierge at scale – a conversational AI assistant covering inspiration, hotel comparison, booking management, in-stay questions and loyalty coaching, available on the ALL Accor website and app and through WhatsApp and iMessage. Accor calls it hospitality’s first end-to-end conversational travel companion deployed at scale. It learned from more than one million guest conversations during a pilot that began in July 2025, and guests can hand over to a human agent on request.
Put this next to this week’s lead story and the strategic logic is obvious. If Meta and Google are building agents that sit above every brand, Accor needs a conversational front door of its own – one that recognises returning guests, knows their loyalty profile and can close the booking without an intermediary. Chief commercial, digital and tech officer Alix Boulnois was explicit that brands now have to be built for AI discovery, and Accor’s earlier move to put the ALL Accor app inside ChatGPT shows it is playing both sides: present in third-party agents, and owning its own.
Marriott, Hilton and IHG all have comparable tools, so the chatbot itself won’t be the differentiator. What sits behind it will. Boulnois’s more interesting remark was about loyalty – that today’s programmes are mostly transactional and reward the past, while AI allows something more predictive and personal. That fits with how Accor is reimagining loyalty economics more broadly. For independents and smaller groups the takeaway isn’t to build a concierge; it’s that direct booking is becoming a conversation, and your website needs to be able to hold one – or at least hand one off cleanly.