Expedia shelved Roamie, its AI chatbot assistant, after discovering that comprehensive end-to-end travel solutions don't yet work reliably through artificial intelligence. CEO Ariane Gorin acknowledged the company's misstep: launching a chatbot that tried to handle every aspect of travel planning from flight bookings to hotel reservations to itinerary creation proved impractical.
The pivot reveals a hard truth about AI in travel. Companies racing to deploy chatbots often overestimate what the technology can accomplish. Expedia's experience shows that current AI struggles with the complexity of actual travel booking, where variables multiply instantly and errors ripple through entire trips. A chatbot might excel at answering FAQs or suggesting destinations, but managing real transactions with real consequences demands human intervention.
The travel industry invested heavily in AI chatbots during the pandemic, expecting them to reduce customer service costs while improving experiences. Airlines, hotel chains, and booking platforms launched versions across their platforms. Many discovered, as Expedia did, that travelers ultimately distrust automated systems for major purchases. When someone books a 7,000-dollar flight or makes a non-refundable hotel commitment, they want human judgment available.
Gorin's candor about Roamie matters because other travel companies face identical challenges. Airbnb, Booking.com, and airline carriers have all deployed conversational AI with mixed results. Some focused on narrower tasks instead. Rather than managing entire bookings, better-performing chatbots handle rebooking after cancellations, provide destination recommendations, or process simple refund requests.
The future of AI in travel remains open. Gorin suggested that end-to-end chatbot solutions might work eventually, once the underlying technology matures and data quality improves. For now, the lesson sticks: test-and-learn matters, but so does knowing when