Inspiration
With Google's extensive suite of APIs like Maps, Places, Hotels, and more, the opportunity to create a unified, intelligent travel experience was irresistible. Gemini 3.0's advanced reasoning and multi-step thinking capabilities make it the perfect foundation for a travel planning system that needs to balance complex constraints, user preferences, and real-world logistics.
What it does
Dora is your AI travel architect. Simply describe your dream trip, and Dora handles the rest
Intelligent Recommendations: Finds optimal flights, accommodations, and activities tailored to your preferences
Smart Itinerary Planning: Crafts day-by-day schedules that balance your interests, budget, and travel style
Interactive Trip Visualization: Brings your journey to life with an immersive map view showing your entire route, stays, and activities
Conversational Interface: Refines your trip through natural dialogue—just say what you want Dora transforms travel planning from a tedious research task into an exciting, personalized experience.
How we built it
Dora is built as an agentic AI system powered by Google's Gemini models. Note : All the models are either gemini-3-pro-preiew or gemini-3-flash-preview
Orchestrator Agent: Routes user messages, manages preference extraction, and coordinates the planning workflow
Deep Itinerary Planner: A reasoning-capable agent that intelligently selects from multiple tools (restaurant finder, weather API, local news) and specialized sub-agents (flights, accommodations, activities) to build comprehensive itineraries
Travel Map Generator: Visualizes the complete journey with interactive markers and routes
Challenges we ran into
Even with a paid tier, Gemini's rate limits became a significant bottleneck during development. The solution required building a custom rate limiter that intelligently queues and throttles requests while maintaining the agentic system's responsiveness—a challenging balance between user experience and API constraints.
Accomplishments that we're proud of
The Travel Map visualization is the feature I've always wanted in AI travel planners but never found. Seeing your entire journey laid out geographically—flights, hotels, activities, and routes—transforms abstract planning into something tangible and exciting. Building this interactive experience from scratch was incredibly rewarding.
What we learned
Rapid iteration doesn't mean wasteful API usage. Building the rate limiter taught me that thoughtful architecture and constraint handling are just as important as feature development—especially when working with powerful but rate-limited AI models. Optimization of resources is a conscious choice.
What's next for Dora - Your travel architect
Phase 1 - Seamless Booking Integration: Direct booking capabilities for flights, accommodations, and activities Price tracking and deal alerts
Phase 2 - Visual Trip Planning: Mood Boards: AI-generated visual collections showcasing destinations, local culture, and experiences Style Boards: Personalized packing and outfit recommendations based on weather, cultural norms, activities, and personal style
Phase 3 - Collaborative Planning: Multi-user trip planning with shared preferences and voting on activities
Built With
- gemini
- google-cloud
- google-maps
- langchain
Log in or sign up for Devpost to join the conversation.