21st of September 2026: Here are a few of the stories that caught our attention over the week, with some of our takeaways.
Story 1
Fliggy’s ‘flyai’ Skill and the Rise of Agentic Booking in China
Hospitality Net’s recent analysis of China’s 2026 Spring Festival travel data makes an argument every distribution professional outside China should read twice: AI has stopped being an inspiration tool for Chinese travellers and has become the transaction engine itself.
The trigger is Fliggy, Alibaba’s travel platform, and its “flyai” skill. Rather than building another app, Fliggy pushed flyai directly into the native skill stores of Huawei, Xiaomi, OPPO and Honor – the operating system layer, not the app layer. A traveller can now book flights, hotels, attractions and ground transport by voice or text without opening Fliggy at all. During Spring Festival, AI-driven orders on the platform rose more than eight-fold on pre-holiday levels, with AI-powered ticket bookings up 24 times.
What makes flyai worth watching commercially, rather than just technically, is where Fliggy has chosen to sit. It has built on its own Model Context Protocol standard, integrated with agent platforms including Kimi, QClaw, MaxClaw and ArkClaw, and requires no account setup – meaning Alibaba captures the guest’s intent at the point of decision, before the guest has chosen which app, let alone which OTA, to use.
Fliggy is not alone in repositioning around agentic AI. Meituan has stood up a dedicated AI Transformation department and is pursuing what its leadership calls a “To A” approach – serving AI agents rather than only guests, opening its fulfilment network so that agents on other platforms, including Tencent’s Yuanbao, can complete bookings through Meituan without the guest switching apps. Trip.com, meanwhile, has restructured its leadership around its TripGenie assistant, which the company says now drives 60% of user interactions tied directly to booking decisions.
The commercial read for international hotels and DMOs is straightforward, even if the fix is not. As Terence Ronson, founder of Pertlink, puts it, any hotel or DMO relationship that assumes a Chinese guest arrives via familiar app search “is now standing on a shrinking assumption.” If agents are choosing where to book based on structured data rather than brand familiarity, then having a clean, machine-readable feed – accurate rates, availability and property data an agent can actually parse – matters more than a listing on Fliggy or Ctrip ever did on its own.
None of this is China-specific for long. The three-way contest between model providers, device makers and service platforms for control of the booking moment is the same contest that will eventually play out everywhere agentic search takes hold. A machine-readability audit, and a genuine view of AI override rates once bookings start completing without a human in the loop, look like sensible groundwork to start now rather than after the pattern repeats outside China.
Story 2
Icelandair’s Answer to AI Search: Build the Brand, Not Just the Prompt
Breaking Travel News carried a striking argument from Gísli Brynjolfsson, Icelandair’s director of global marketing, this week: the fight for visibility in AI-generated travel recommendations is won or lost before the traveller ever types a prompt.
His logic runs against the instinct many of us have built up over fifteen years of SEO and metasearch optimisation. Brynjolfsson argues that because nobody yet fully understands how large language models decide which brands to recommend, the only reliable lever left is brand itself – something travellers already know, remember and feel something about before they ever ask an AI for advice. Optimise for the algorithm today and you are optimising for a system that may work completely differently in six months. Build a brand people recognise, and you are hedged against that uncertainty regardless of how the models change underneath it.
Icelandair’s response has been to lean deliberately into human authenticity rather than polish. With AI-generated travel content proliferating – and travellers reportedly growing weary of it – the airline has built campaigns around real staff, real routes and unscripted moments rather than the airbrushed imagery that increasingly reads as synthetic even when it isn’t. The bet is that in a market flooded with AI content, looking obviously human becomes a point of difference rather than a limitation, and results so far are apparently encouraging enough that the airline is doubling down rather than pulling back.
None of this means Icelandair has stepped back from data. The airline still tracks search behaviour and website activity, booking history, and social listening to understand demand and identify audiences. Brynjolfsson’s point is narrower, and for a commercial audience more useful: technology should sharpen the airline’s understanding of people, not replace the judgement about what to say to them. As he puts it, technology should make marketing more human, not less.
For hotel commercial teams watching the same AI-search shift reshape how guests discover properties, the practical takeaway has less to do with chasing this month’s prompt-engineering trick and more to do with long-term brand investment – the kind of recognition that survives whichever model or interface travellers happen to be using next year. Brynjolfsson is due to expand on the thinking at the World Aviation Festival’s Marketing Summit in Lisbon this October, and it will be worth watching whether the argument holds up once more airlines and hotel groups test it against real booking data rather than theory alone.
Story 3
Europe’s Hotels Still Lead on Direct Bookings, But the AI Gap Is Opening
PhocusWire’s write-up of HOTREC’s European Hotel Distribution Study 2026 – based on responses from 2,713 properties – is a useful corrective for anyone assuming direct-versus-OTA is a settled argument. European hotels do still hold a narrow majority of bookings direct, at 51%, but OTA share has grown to 29.9%, up from 19% in 2013, and that growth is concentrating rather than spreading.
Booking Holdings and Expedia Group between them control more than 85% of European OTA bookings, with Booking Holdings alone accounting for close to 69% of the OTA-mediated total. Booking Holdings was formally designated a “gatekeeper” under the EU’s Digital Markets Act in 2024, and HOTREC is now pushing regulators to enforce that designation properly, alongside the transparency protections in the Platform-to-Business regulation, which HOTREC fears could be diluted by the European Commission’s red-tape-cutting “Digital Omnibus” initiative.
The friction hotels report is concrete rather than abstract. Just over half – 51% – say OTAs undercut their own direct rates frequently or occasionally, up eight points since 2023. Nearly 44% report their rates or availability being redistributed by third parties without their involvement, producing the familiar mess of price inconsistency, guest confusion and lost inventory control. HOTREC’s president, Alexandros Vassilikos, frames the ask sensibly: hotels are not looking to walk away from digital platforms, only to keep genuine choice and a competitive marketplace rather than one-way dependency.
The AI section of the study is where it gets interesting for anyone thinking about the next three years rather than the last three. Only 26% of European hotels have adopted AI so far, and adoption skews hard towards larger chains, urban properties and higher-category hotels – exactly the segment with the resources to move first. The direction of travel is from AI as a research aid towards agentic booking, where an AI agent completes the transaction on a traveller’s behalf. Research from Amadeus cited in the piece suggests travellers want the personalisation AI offers but not unchecked autonomy – a human in the loop still matters for anything beyond routine servicing.
For independent and mid-market hotels reading this from outside the 26%, the gap is worth taking seriously. Regulatory pressure on Booking and Expedia may eventually rebalance the OTA relationship, but it will not, on its own, close the AI adoption gap opening up between large chains and everyone else. That is a competitive disadvantage building quietly while attention stays fixed on commission rates and rate parity.
Story 4
Loews Bets on Approachable Wellness With Active Stays
Loews Hotels’ portfolio-wide Active Stays programme is a useful case study in how a hotel group can chase the wellness travel trend without turning every property into a boot camp. Industry survey data cited alongside the launch puts real weight behind the shift: 60% of active travellers now rate fitness amenities as highly as any other hotel perk, which moves wellness from a nice-to-have into a genuine booking driver. That is not a marginal shift – it means a hotel’s gym and its restaurant menu are now competing with its view and its rate for a guest’s attention when they are choosing where to stay.
Where Active Stays differs from the approach several competitors have taken is in its restraint. Properties such as 1 Hotel Hanalei Bay and Hotel Commonwealth have built their wellness positioning around performance – serious training equipment, structured programmes, measurable targets. Loews has gone the other way, building the initiative around flexible, low-commitment movement that fits around a guest’s own schedule rather than around a plan. In partnership with fitness expert Makena Rae, Loews has put a library of equipment-free, bodyweight workouts on guest room televisions, so a guest can do ten minutes of mobility work without ever booking gym time.
The more interesting commercial decision is the localisation. Rather than rolling out one national programme, individual properties have built their own version around what is actually on their doorstep. Loews Regency New York offers a Central Park running package; Loews Chicago has mapped a self-guided route along Route 66 to Navy Pier; Loews Kansas City has built its offer around an indoor saltwater pool and Peloton bikes; Loews Ventana Canyon in Tucson leans on its desert setting with trail shuttles and sunrise yoga. None of it requires new capital-heavy infrastructure at group level – it is largely a case of packaging what a property already has, or a light-touch local partnership, around one shared brand idea rather than commissioning something new for every hotel.
The programme also extends past the workout itself into recovery and food, under a “move well, recover well” framing – hydration-led drinks, protein-forward dishes and grab-and-go recovery options across restaurants and in-room dining menus, so the wellness positioning does not stop the moment the guest leaves the gym floor.
For commercial teams weighing whether wellness is worth a dedicated push, Active Stays is a reasonable template for how to do it without a large capital commitment: borrow an existing local asset, add one signature partnership, and let each property’s own version of the story do the marketing work rather than relying on a single, centrally produced campaign.
It’s hard not to read Active Stays as a live example of the thinking behind BrandMatch, the wellness and lifestyle marketplace Mario Chessa and I are building for an October launch. Loews isn’t carrying spa build-outs or trainers on payroll at every property – it’s licensing one relationship with Makena Rae and letting each hotel plug in whichever local wellness partner already suits its market. That’s the ancillary-revenue logic our TRevPAG framework is built around: value attached to the stay through partnership, not added to next year’s capital budget.
Story 5
Anana Wants to Capture the Guest Intent Your PMS Never Sees
Hospitality Net’s coverage of Anana’s launch points at a genuine gap in most hotel technology stacks: the systems built to record a booking are not the same systems that ever captured why the guest wanted it. Founded by Ricardo Pantaleón and Alexandros Zisimidis and aimed squarely at commercial teams – CCOs, DOSMs, CROs and CMOs – Anana is positioned as a workspace that pulls together the conversational signal a PMS or CRM was never designed to hold.
The problem it is solving is familiar to anyone who has sat through a revenue meeting trying to explain a demand shift after the fact. A guest asks about connecting rooms, mentions mobility needs, or references a celebration on a call, a chat or an email, and once that interaction ends the detail is usually gone – it never made it into a structured field anywhere. Multiply that across every voice call, live chat and AI discovery engine a hotel now touches, including ChatGPT and Perplexity as sources of guest intent, and the scale of what is being lost becomes clear.
Anana’s method, which it calls a “Hospitality Context Protocol,” pulls from PMS, CRM, telephony and even external inputs like weather and local events, re-weaving the picture roughly every fifteen minutes. It looks for agreement across three or more sources before it treats something as a genuine pattern – a rise in pickup alongside a cluster of guest messages about a specific weekend, for instance, gets flagged as a demand shift with a recommendation attached, rather than just another alert to ignore. One case study in the coverage has the system spotting a cluster of spa staffing complaints concentrated after 4pm on weekends, then costing out a fix – an additional therapist at $1,820 – against an estimated Net Promoter Score recovery, giving operations a number to act on rather than a vague sense that service was slipping.
The founding team’s decision to handle implementation directly, rather than leaving it to a generic onboarding process, suggests they understand that a tool like this lives or dies on how cleanly it sits alongside a hotel’s existing PMS, CRM and telephony rather than operating as a bolt-on dashboard.
Every conclusion the system reaches is traceable back to its source call, email or document, which matters for a commercial team that will need to defend a rate change or a staffing decision to ownership. Whether Anana can prove that value at portfolio scale, beyond the single-property case studies it is launching with, is the question worth watching over the next few quarters.