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

The trains have been booked, the operator thinks it is done. But I need to be somewhere I remind it – can it book a hotel? It asks more details and I am purposefully obscure, mentioning that it should be ease and conveniently located. Compared to hotels is probably the least favorite aspect of my travel plan, so I am happy to leave it scrolling it via booking.com. When I see it sets the wrong dates, I keep myself from the barrier, but it itself corrects it. It spends a while to survey the list of IBIS, but finish choosing a three-star hotel called Martin’s Brug, which I notice that users have rated as a great position.

What is left now is a journey. Here, the operator seems to lose steam. It provides a Perfectry ODI schedule that originally seems to be crushed from the vegetarian travel blog. On the second day, it suggests that “I look at any remaining attractions or museums.” Wow, thanks for the tip.

The day of the trip arrives, and when I pull myself out of bed at four o’clock in the morning, I think why I usually avoid the initial exit. Nevertheless, I reached Brussels without issue. My ticket is allowed to travel but I understand that I don’t know where I am going. I have shot the operator on my phone and ask which platform to leave the next Bruce-Band train. It searches for Belgium railway schedules. A few minutes later, it is still searching. I look and see the details on the station display. I reached the platform before the operator took it out.

Bruces are pleasant. Given the operator’s unnecessary travel route, I branch out. This type of research task is suitable for a large language model, I understand – it does not require agent power. The operator’s OpenAI Siblings Chatzept gives me many more thorough plans, not only where to eat, but what to order (De Hallov Man Bruwari Flemish Stew) conspires with the advice of the moment. I also try Google And the ethnographic clode and their plans are the same: walk on the market squares; See the Balfree Tower; See the Basilica of the Holy Blood. Broze is a small town, and I can’t help but I can’t imagine that it is only standard tourist route, or if the AI ​​models are all getting their information from the same source.

Various travel-specific AI tools are trying to break through this genericity. I briefly try the MindTrip, which provides a map in addition to a writteny route, proposes to personalize recommendations based on a quiz, and it includes collaborative features for shared travel. CEO Andy Moss says that it extends to the Broad LLM capacity through the benefits of travel-specific “knowledge base” with things like weather data and real-time availability.

Victoria courtesy of Turk