Trip Planner
A travel planner built on the premise that for decisions that matter, AI should help you see the whole field, not hand you a shortlist.
Designer & developer
Personal project
Problem
The promise of AI planning tools has been: let us shortcut this for you. Describe what you want, get a handful of recommendations, pick one, move on. But for decisions that matter, like where to stay in a place you've never been or how to spend a week you can't redo, the shortcut is the thing that breaks trust. NN/g's research puts a name on it: people maximize for important decisions. They want to see the full field before choosing. They want to know what got filtered out. A shortlist with no visible rationale feels less trustworthy than a plain search, not more. So the question isn't whether AI can plan a trip. It's how AI helps without sending you down a path you can't verify.
Approach
I built the planner around a different premise: show everything, make the ranking visible, let the user's reactions do the narrowing. There are two paths through the app. Explore, where you're building the set of places worth considering. And Plan, where you turn that set into a real itinerary.
Explore is how places come in. A single Add Place drawer accepts whatever format the recommendation actually arrives in: a Google Maps link from a friend, a screenshot of an Instagram post, a blog post with 15 picks, a messy list pasted from Notes, a place name typed into the field, or a region lassoed on the map. Each input type streams places into the same results view. Google Places fills in photos, ratings, review counts, and hours under the surface, so every card carries real social proof before any AI opinion enters the frame. A lightweight profile learns from what you rate, save, and skip, and quietly reshapes the ranking as you go.
Plan is where the ranked set becomes a trip. You cluster places by geography, reorder days by dragging, and iterate (moving things between days, adding and removing, swapping stops) until the shape of the trip feels right. A homebase toggles the model: out-and-back loops with walking, biking, or driving modes when you're staying put, forward-progress routing when you're moving between stops. Google Directions fills in real travel times, so a five-hour driving day looks like one before you commit.
Rate & enrich imported places
Research by highlighting areas
Plan mode — road trip
Plan mode — with a homebase
Impact
The planner takes on the parts of trip planning that get unwieldy fast: collecting options from every direction they arrive in, stitching them into something comparable, keeping track of what you've reacted to, and reshaping a day as it grows or shrinks. AI is doing the fetching, summarizing, and estimating in the background, like photos for the cards, travel times between stops, context you'd otherwise open fifteen tabs for. The planning itself stays with you. The result is a trip you actually chose, built on information you can check.
