I Let AI Agents Plan My Vacation—and It Wasn’t Terrible

Reserved trains, the operator thinks that his work is finished. But I will need a place to stay, can I call it back – can it book a hotel? He asks for more details and I am deliberately vague, specifying that he must be comfortable and ideally located. Comparing hotels may be my least favorite appearance of travel planning, so I’m happy to let it scroll via Booking.com. I withdraw myself from jumping when I see that the bad dates are defined, but that corrects it itself. He spends some time investigating an Ibis list, but ends up choosing a three -star hotel called Martin’s Brugge, which I note that users have evaluated as having an excellent location.

Now there is only one route remains. Here, the operator seems to lose steam. It offers a surface day schedule which seems to have been mainly led from a vegetarian travel blog. Day 2, it suggests “visiting the remaining attractions or museums”. Wow, thanks for the advice.

The day of the trip arrives and, while I hang out from the bed at 4:30 am, I remember why I generally avoid early departures. However, I arrive in Brussels without problem. My ticket allows you to travel, but I realize that I don’t know where I’m going. I am launching an operator on my phone and I ask which platform the next train to Bruges leaves. He is looking for the schedules for Belgian railways. A few minutes later, he is still looking. I look up and see the details on a station screen. I arrive at the platform before the operator understands it.

Bruges is delicious. Given the operator’s dull route, I diversify. This type of search task is perfect for a wide language model, I realize it – it does not require agenic capacities. Chatgpt, the operator Openai Sibling, gives me a much more in -depth plan, plotting activities on time with suggestions not only to eat, but what to order (Flemish stew at the Halve Mann Brewery). I also try Google Gemini And the Claude d’Anthropic, and their plans are similar: walking to the market place; See the tower of belfry; Visit the basilica of sacred blood. Bruges is a small town, and I can’t help but ask myself if it is simply the standard tourist route, or if AI models get all their information from the same sources.

Various AI -specific AI tools try to unravel this credits. I briefly try Mindtrip, who provides a card alongside a written route, proposes to personalize recommendations based on a quiz and includes collaborative features for shared trips. The CEO Andy Moss says that he develops large LLM capacities by taking advantage of a “knowledge base” specific to travel containing things such as weather data and real -time availability.

With the kind permission of Victoria Turk