AI travel assistants surged over Golden Week, but booking execution stalls
Over China's National Day holiday, platform AI travel assistants posted sharp usage gains — Fliggy Bangbang interactions rose 158% month-on-month with AI-guided flight and hotel bookings up 110%, Meituan's AI assistant "Xiaotuan" saw users and calls up more than 60% versus May Day, and Tongcheng's DeepTrip drew over 3x hourly visits at its first-day peak, per platform figures cited in the report. Media tests nonetheless found AI itineraries recommending sold-out trains, wrong metro transfers and attractions closed on the day of visit. The article attributes the gap to delayed third-party data (GDS flight caches refreshing every 2–3 minutes, hotel inventory lags of 10–15 minutes), fragmented airline/hotel/scenic-spot data, and AI's lack of payment authority, with only 17% of surveyed users willing to let AI handle payments or passport details unsupervised.
Travel is one of the earliest mass consumer use cases for chat assistants, so fixing data freshness and payment permissions will decide whether AI can complete transactions rather than just suggest them.