Fliggy's 800% Booking Surge Highlights AI's Distribution Advantage

The Big Picture

Fliggy's booking assistant now ships inside Huawei and Xiaomi phones. Bookings through it rose 800% during Spring Festival. No app download, no search page, no hotel website.

Meanwhile half the hotels in the West are running chatbot pilots and calling it transformation.

That's the thread this week. The gap between AI adoption and AI impact isn't a technology gap. It's a position gap. Chinese platforms embedded themselves at the operating system layer, where intent forms. Everyone else bought software that sits behind the booking, optimising work that was already happening. Same models. Completely different leverage.

Brief of the Week

Over half of hotels now run some form of AI. Fewer than one in ten report real impact. That gap, from State of Distribution 2026, ends the adoption debate and opens a harder one: why no results.

The Number

800%

According to Hospitality Net, Fliggy's AI booking feature drove an 800% increase in hotel rooms booked during Spring Festival after being integrated into major Chinese smartphone brands. This means the interface layer, not the inventory layer, is where distribution power is moving.

01

Fliggy proves the OS is the new OTA

According to Hospitality Net, Fliggy's assistant is now native to Huawei and Xiaomi devices, and Spring Festival bookings through it grew 800%. The lesson for European and US operators: your commercial relationship with the guest can be intercepted before any browser opens. The same report notes the US RevPAR growth streak ended after 21 weeks, which means less pricing cushion while this shift happens.

02

Half adopt, under one in ten benefit

According to the State of Distribution 2026 report covered by Hospitality Net, over 50% of hotels use AI but fewer than 10% report meaningful impact. This means most AI budgets are funding internal efficiency, not demand capture. OTAs still dominate the channel mix, and the report makes a second point worth sitting with: your guest reviews are now read primarily by machines, not people.

03

Mindtrip raises $22.5M and goes after DMOs

According to PhocusWire, Mindtrip closed $22.5M with Capital One Ventures and United Airlines Ventures, launched Mindtrip Stays for hotel search and booking, and is now working with 35+ DMOs in Europe, including Madeira and Norway. This means destinations are buying agentic distribution before most hotels have finished writing their AI policy.

04

OECD sets the policy floor

The OECD's September policy paper on AI and tourism in APEC economies identifies six priority areas for responsible adoption, with specific attention to SME dominated tourism sectors. This means the regulatory conversation has moved from "should we" to "under what conditions", and small operators are being named as the structural risk, correctly.

05

ETIAS and fuel, the boring constraints

According to Hosteltur, jet fuel jumped 18% in two weeks to $4.58 per gallon on Middle East tensions, and airlines still lack a reliable way to verify ETIAS authorisations arriving as emails and QR codes. This means margin pressure and verification friction will hit operations well before any agentic booking flow does.

Contrarian Take

Stop asking whether your hotel or destination uses AI. Wrong question.

Flip the camera. A traveller in Milan opens their phone, says where they want to go, and an assistant returns three options with prices. Your property either appears in that answer or it does not exist. Nothing in your PMS, your revenue tool, or your internal copilot changes that outcome. What changes it is whether your rates, availability, policies, and content are machine readable and reachable through an API a model can actually call.

Most AI spending in hospitality right now is defensive automation dressed as strategy. The 10% seeing impact are the ones who fixed their data plumbing first.

Where This Goes

Skift Global Forum opens in New York tomorrow and you will hear plenty about agents. Ignore the demos. Before winter booking windows open, audit one thing: can an external model retrieve your live inventory and book it without a human? If the answer is no, that is your entire 2027 roadmap.