I am not someone who enjoys spending weeks planning a trip. I want to go places. The research, the spreadsheet of hotels, the itinerary that needs to account for six different opinions — that part I find genuinely draining.
This year I decided to see how much of it I could hand off to AI. The answer was: more than I expected.
What I Used AI For
The first thing I did was give Claude a detailed brief. Destination, travel dates, number of people, rough budget, what we actually want from a trip (not just “culture and food” — specifically what kind of experience, pace, and vibe). I asked it to build an initial itinerary that accounted for all of that.
What I got back was substantially more useful than anything I could have put together in an equivalent amount of time. A day-by-day structure that made logical geographic sense, so we were not doubling back across a city unnecessarily. Restaurant suggestions organized by meal type and neighborhood. A realistic sense of how long different activities actually take, rather than the overly optimistic “you can do this in an hour” estimates you often see in travel guides.
I pushed back on parts of it, asked for alternatives, changed the priorities as we talked — and the itinerary got better with each iteration. It felt like working with a well-traveled friend who had done the research rather than consulting a generic guidebook.
Where Perplexity Came In
For anything that required current information — whether specific restaurants were still open, what the current entry requirements were, recent traveler reports about specific neighborhoods — I switched to Perplexity. Language models have knowledge cutoffs, and for travel planning the gap between “what was true a year ago” and “what is true now” can matter significantly.
Perplexity pulled current information and cited the sources, so I could see whether I was reading something recent or something from two years ago. For visa requirements and travel advisories especially, that recency matters.
What AI Did Not Do Well
Specific booking. AI can tell you which areas to look for hotels in and what kind of property suits your travel style. It cannot actually check availability and prices in real time. For that I still used the normal booking sites. It narrowed down what I was looking for significantly, which made the actual booking faster, but the final step was still manual.
Hyper-local hidden gems. AI knows what is well-documented online. The genuinely off-the-beaten-path recommendations — the places that do not have a lot of English-language coverage yet — are still better found through people who have actually been there recently. For those I turned to travel forums and friends who knew the destination.
The unexpected. No amount of planning accounts for a festival happening that weekend, a recommended restaurant being fully booked for a week, or the fact that you are going to fall in love with a neighborhood you had allocated an hour to and want to spend the whole day there. AI builds a plan. Travel is what happens when the plan meets reality.
The Overall Verdict
Using AI for trip planning did not make the trip better — the trip was good because of the destination and the people I was with. What it did was make the planning significantly less painful and genuinely faster.
The research phase that used to take me the better part of a weekend took most of an afternoon. The itinerary I ended up with was more coherent than ones I have assembled manually, because AI thinks about the whole thing holistically rather than building it piece by piece as I happened to find good things.
Would I use it again? Already planning the next trip the same way. The process has changed enough that I would not go back.