Tripemini

Step 1
Step 2
Output 1/2
Output 2/2

Inspiration

Most travel planning tools start with destinations, budgets, or schedules. But for many people, trips actually start with taste. Food is often the strongest signal of personal preference, yet it's rarely used as a primary input for planning an entire trip.

We wanted to explore: What if a single photo of food you love could determine where you go, what you eat, and how your days are organized?

What it does

Tripemini is an AI-powered travel planner that generates complete itineraries based on your taste. Upload food photos you love, enter your destination and dates, and get a personalized trip with:

Restaurants matched to your cuisine and flavor preferences
Nearby attractions organized into optimized daily routes
Budget estimates for flights, hotels, food, and activities
Explanations for why each place was selected

How we built it

Tripemini uses Gemini 3 Pro Preview for multimodal reasoning:

Gemini Vision analyzes food images to extract cuisine and taste signals
Gemini Reasoning balances constraints like distance, ratings, and timing
Next.js 16 powers the responsive web app with Tailwind CSS and Zustand

Challenges we ran into

Translating subjective food preferences into structured, searchable signals
Coordinating real-world constraints like distance, time, and availability
Ensuring itineraries are coherent and realistic, not just random lists

Accomplishments that we're proud of

A novel multimodal entry point for travel planning
Fully automated itinerary generation from minimal user input
Clear "why selected" explanations for every recommendation

What we learned

Multimodal inputs significantly improve personalization. Starting from images rather than text allows users to express intent more naturally. Combining search with reasoning is essential for real-world planning.