How we plan a trip in 48 hours

Sofia and I used to spend weeks planning a weekend trip. Tabs everywhere. A Notion page nobody opened twice. Flights we “almost” booked three times before we lost the good price. The whole thing felt like homework for a vacation we didn’t want anymore.

Now we do it in 48 hours. One spreadsheet, a few AI tools, and rules we actually follow. Here’s the system.

The spreadsheet

One tab per trip. Columns: day, morning, afternoon, evening, reservation status, walking distance, backup option. That’s it. No color coding. No dropdowns. No “categories.”

The backup column is the one that saves us. For every reservation-required thing, there’s a plan B that doesn’t need one. Restaurant closed? We walk ten minutes to the tascas around it. Museum line too long? We swap the afternoon with Thursday’s. The trip stays intact.

The AI tools we use

ChatGPT for shortlist generation. The prompt is boring and works: “We have 48 hours in [city], staying in [neighborhood]. We like [specifics]. Give us a list of 12 restaurants ranked by how memorable they are, with one line explaining why, and tag which need reservations.”

Claude for the itinerary draft. Paste the shortlist, add a few constraints (“Sofia needs a nap after lunch, we hate waking up early on Sundays”), and ask for a 48-hour schedule with walking routes grouped by neighborhood. It does the geometry you’d otherwise spend an hour doing in Google Maps.

Perplexity for fact-checking hours and reservations. AI tools lie about business hours. Perplexity pulls from current sources. We cross-check every reservation before we call, every time.

The rules we follow

No more than three “must do” items per day. Anything above three is a commitment, not a trip. If one goes sideways, the whole day buckles.

Every day has a slow window. Two hours minimum with nothing scheduled. This is when you find the trip’s actual memories; the bookshop you wander into, the park bench, the barista who remembers your order on day two.

One splurge. One. Either a dinner, a hotel upgrade, or a private thing (a driver, a tour, a cooking class). Not all three. We used to try to maximize every category and came home broke and tired. Now we pick the one that matches the trip’s mood.

What we skip

Travel blogs with 400-item lists. The algorithm rewards length, not taste. Ten good recommendations from a person with taste beat a hundred-item list every time.

Pre-booking every meal. Our best dinners are always the ones we found at 3pm by walking by and peeking at the chalkboard. Book the two places you’d regret missing. Leave the rest to the day.

Packing cubes. Sorry. We tried. They’re a cult. A duffel and a laundry bag are a travel system.

The 48-hour timeline

Friday night: we write a one-line intent. “What are we actually looking for?” It’s usually food, quiet, and one story worth telling. This line anchors the whole plan.

Saturday morning: the spreadsheet. Flights shortlisted, hotel booked, shortlist of activities generated. We aim to spend two hours max here.

Saturday afternoon: draft itinerary in Claude, cross-check in Perplexity, refine in the spreadsheet. Another two hours.

Sunday morning: reservations. All of them. In one sitting. If a restaurant doesn’t take online bookings, we add it to the list and Sofia calls on her lunch break Monday.

Sunday night: flights booked. Bags mentally packed. The trip exists.

Why this works

The planning isn’t the fun part. The trip is. Spending six weekends researching a three-day trip is how you kill the anticipation and arrive already tired. Doing it in 48 hours forces good enough decisions, which turn out to be the right ones 95% of the time.

And the 5% you get wrong? That’s the story you tell at dinner.