Operator AI struggles with travel itineraries despite booking trains and hotels smoothly.
The AI handled train tickets and hotel bookings with ease—fixing errors and picking a well-rated three-star hotel in Bruges. But its itinerary? Weak. Day one was basic, mostly lifted from a veggie travel blog. Day two? A lazy “visit any remaining attractions or museums.”
At the station, Operator lagged too. It couldn’t quickly find the platform for the next Bruges-bound train, stalling while human users grabbed the info from displays. 
Bruges itself was charming. But Operator’s itinerary flopped. ChatGPT and other LLMs like Google’s Gemini and Anthropic’s Claude delivered better hour-by-hour plans, with dining recs and key stops like the belfry and the Basilica of the Holy Blood. All seemed to pull from the same generic sources, though.
Specialized travel AIs try to fix this. MindTrip, for example, adds maps, personalized quizzes, and real-time data to LLM smarts. CEO Andy Moss says his product builds on broad AI by pulling in travel-specific info like weather and availability.
Operator AI still has a long way to go before it handles trips end-to-end.
Courtesy of Victoria Turk
