Triplix - An AI based Travel planner App

AI-Generated Itinerary — Complete day-by-day travel plan with timings, attractions, and local recommendations
Real-time Budget Tracking — Track spending across all categories with per-person cost splitting for group travel
Swipe to Discover — Tinder-style interface for browsing hotels, destinations, and attractions with AI-powered recommendations
Trip Reel Creator — Automatically generate shareable travel reels from your trip photos using AI
AI Travel Assistant — Natural language chat interface powered by Gemini 2.5 Flash for end-to-end trip planning
ransport Search — Compare flights, trains, buses, and taxis with real-time prices and booking options
Where To Next — AI-curated destination suggestions based on user preferences, budget, and travel style
Personalized Preferences — Users set dietary needs, accessibility requirements, and special requests for a tailored travel experience
Real-time Budget Tracking — Track spending across all categories with per-person cost splitting for group travel
Data Flow Design — User request flows through 8 specialized AI agents for hotels, transport, itinerary, and budget coordination
System Architecture — Flutter frontend connected to FastAPI backend with 8 Google ADK agents powered by Gemini 2.5 Flash

Inspiration

Planning a trip in India is overwhelming — dozens of booking platforms, hundreds of options, no single app that understands you. We were inspired by how dating apps (Tinder) make decisions fun and effortless through swiping. What if planning a trip felt the same way? What if an AI could understand your preferences, curate options, and build your perfect itinerary — all through natural conversation and intuitive swipes?

That's how Triplix was born.

What it does

Triplix is an AI-powered travel planning assistant that combines conversational AI with a swipe-based interface to make trip planning effortless:

AI Chat Interface — Talk naturally: "Plan a 5-day trip from Delhi to Goa for 3 people under ₹50,000." The AI understands context, remembers preferences, and builds your trip step-by-step.
Tinder-Style Swipe Cards — Browse flights, trains, hotels, and destinations by swiping right (accept) or left (skip). The AI learns your taste and refines suggestions.
Multi-Agent Architecture — 16 specialized AI agents work together:
- Manager Agent orchestrates the workflow
- Transport Agent finds flights, trains, buses, taxis
- Hotel Agent searches accommodations with real Google Maps data
- Destination Agent recommends places with city-specific attractions
- Itinerary Agent creates day-by-day plans with weather forecasts
- Budget Tracker Agent manages expenses and group splits
- Preferences Analyzer learns from your swipe patterns
Smart Budget Manager — AI-powered chatbot for expense tracking. Say "Spent ₹5000 on hotel" and it auto-categorizes, tracks against your budget, and provides group split calculations.
AI Trip Reel — Upload or capture trip photos. Gemini Vision AI automatically filters out documents, receipts, blurry images, and keeps only the best travel moments. Play them as an Instagram Stories-style highlight reel with AI-generated captions.
QR Trip Share — Generate a scannable QR code of your itinerary and share it instantly via WhatsApp, email, or any platform.
Route Maps — Ask for directions and see real Google Maps routes with polylines, distance, duration, and turn-by-turn navigation.
Voice Input — Speak your travel plans using Google Speech-to-Text.

How we built it

Frontend: Flutter (Dart) — runs on web, Android, iOS from a single codebase. 28+ screens with Provider state management, card swiper for Tinder-like UX, and animated transitions.

Backend: Python FastAPI server with 16 Google ADK (Agent Development Kit) sub-agents. Each agent is a specialist with its own tools and Gemini 2.5 Flash model instance.

AI Engine: Google Vertex AI + Gemini 2.5 Flash for:

Conversational trip planning (multi-turn context)
Photo analysis & filtering (Gemini Vision)
Budget intelligence & expense categorization
Itinerary generation with weather integration
Preference learning from swipe patterns

Data Layer: CSV datasets for Indian travel data (flights, trains, hotels, destinations across 30+ cities) with MongoDB Atlas as optional cloud database.