Start with the trip you actually want
A vague prompt produces a vague itinerary. "Plan three days in Lisbon" usually returns the same famous stops in a reasonable order, but it does not know whether you want slow mornings, late dinners, kid-friendly pacing, a low budget, or a trip built around food. The missing context becomes extra editing work.
A stronger prompt begins with the kind of trip you are trying to have. Name the destination, dates or length, travel style, budget level, group makeup, and the pace you prefer. If you hate early mornings or want one anchor activity per day, say that up front.
Give constraints before preferences
Constraints are the facts the itinerary must respect. Arrival time, departure time, hotel neighborhood, mobility needs, dietary restrictions, fixed reservations, and must-skip activities belong near the top. Preferences are softer: food markets, architecture, beaches, live music, bookstores, scenic walks.
Putting constraints first helps the AI avoid plans that look polished but fail in real life. A day that assumes you can start at 8am is not useful if your flight lands at noon. A route that ignores your hotel location may waste the best part of the morning getting across town.
Ask for day shapes, not minute-by-minute control
Minute-by-minute AI itineraries often feel precise without being practical. Travel has too much friction for that level of confidence: lines, transit delays, tired people, weather, and the simple fact that good stops sometimes take longer than expected.
Ask for morning, afternoon, and evening blocks instead. This gives you a useful structure while leaving room to adjust. If a block needs a reservation or has limited hours, ask the AI to call that out separately instead of hiding it inside a dense schedule.
Request tradeoffs and assumptions
A good prompt asks the AI to explain its choices. For example: "Give me a balanced three-day plan, then list the assumptions you made and two tradeoffs." That response is easier to edit because you can see what the itinerary optimized for.
Tradeoffs are where the useful planning happens. Did the plan choose fewer neighborhoods to reduce transit? Did it skip a famous sight because it does not fit the pace? Did it cluster activities by area? These explanations help you decide what to keep.
Use the first draft as raw material
The point of an AI itinerary is not to accept the first answer. It is to get a first pass quickly enough that you can spend your time improving the plan instead of staring at a blank page. Move things around, remove filler, add personal picks, and check anything that depends on hours, tickets, or local conditions.
Salida is built around that editing step. AI can generate a day, but the trip still lives in a normal itinerary where you can adjust pacing, add notes, watch the weather, and keep the plan readable once you are actually traveling.