Operator is satisfied with its task. But I’ll need somewhere to stay, I remind it—can it book a hotel? Then it wants more information, and I give vague answers. For example, that I want a hotel to be convenient and comfortable. Booking.com has become my favorite site to browse through when I want to compare hotels. When I notice that it has the dates wrong, I don’t jump in. It corrects itself. After a long search of the Ibis listings, it chooses a 3-star hotel in Brugge that users have rated highly.
All that is left now is an itinerary. Operator appears to be losing steam. The one-day itinerary is a stale copy of a travel blog for vegetarians. It suggests that I should be eaten on day 2. “visit any remaining attractions or museums.” Thanks for sharing this tip.
When the day arrives for my trip, I wake up at 4:30AM and remember why I normally avoid early departures. Despite this, I make it to Brussels with no problems. But I can’t remember where I want to go. Operator is my go-to app. I ask Operator which platform departs the next Bruges bound train. The Belgian rail timetables are searched. After a few minutes, the search continues. The details are displayed in the display of a train station. Before the operator can figure it out, I reach the platform.
Bruges has a lot to offer. Operator’s itinerary being so mediocre I take a different route. This kind of research task is perfect for a large language model, I realize—it doesn’t require agentic capabilities. ChatGPT (Operator’s OpenAI sister) gives me a more detailed plan. It plots activities hourly and suggests what I should order at restaurants, such as the Flemish stew served by De Halve Mann Brewery. Then I tried Google’s Gemini Anthropic’s Claude and his plans are very similar. They both plan to walk around the square, see the bell tower and visit the Basilica of the Holy Blood. I wonder if these AI models all get the same information, because Bruges, a relatively small city, is based on the usual tourist routes.
Diverse AI-based travel tools attempt to get past this genericity. MindTrip is a travel-specific AI tool that offers a map with a written itinerary. It also provides personalized recommendations using a quiz and has collaborative features to allow for shared trips. Andy Moss of CEO Andy Moss said that the travel-specific LLM expands upon broad LLM abilities. “knowledge base” It can include things such as weather data or real-time accessibility.
Victoria Turk
