System Architecture Diagram

Dora - AI Travel Buddy 🗺️

Inspiration

Travelling should feel exciting, but in reality it often becomes stressful and fragmented. Tourists constantly switch between multiple apps just to navigate a single day: maps for transport, translators for menus, social media for recommendations, and travel apps for planning.

This becomes even harder in countries with language and cultural barriers. Travelers struggle to understand menus, navigate unfamiliar transport systems, verify viral recommendations, or discover places suited to their dietary needs and preferences.

We built Dora to become a single AI-powered travel companion that reduces friction and helps people explore confidently, authentically, and with less stress.

What it does

Dora is an AI-powered travel companion that transforms how people explore new places.

By combining trip planning, local recommendations, transport support, translation, and cultural understanding into one intelligent assistant, Dora helps travellers navigate unfamiliar countries with confidence and ease.

Key Features

🍜 Smart Food Discovery + AI Menu Understanding

Recommends food based on dietary restrictions, budget, cravings, location, and local popularity
Uses image understanding to:
- Translate menus
- Explain unfamiliar ingredients
- Detect dietary concerns
- Provide cultural context

Impact: Users can confidently enjoy local food without language or dietary uncertainty.

📱 Fact-Checked Social Media Recommendations

Instead of blindly trusting viral content:

Travel recommendations are extracted from social media sources
Cross-validated against reviews and ratings
AI summarizes reliability and sentiment

Impact: Users avoid overrated tourist traps and receive more trustworthy suggestions.

🧭 Personalized Exploration + Simplified Transport Guidance

Dora recommends:

Trending attractions
Hidden gems
Local experiences
Seasonal destinations

For transport:

Converts complex route instructions into human-friendly guidance.

Example:

Instead of: “Transfer to JR Yamanote Line via Exit B2”

Dora explains:

“Take the green Yamanote Line from Platform 4 for 3 stops, then follow signs to Hachiko Exit.”

Impact: Makes public transport easier for tourists unfamiliar with local systems.

🌏 Cultural Context + Translation Assistant

Real-time text and sign translation
Cultural etiquette explanations
Context-aware language guidance
Landmark and surroundings interpretation

Impact: Travelers interact more respectfully and understand places beyond literal translations.

How we built it

Frontend

React + Vite
TailwindCSS

Backend & AI Orchestration

Python FastAPI
n8n workflow automation

AI Models

OpenAI → conversational generation + summarization
Reka → vision, translation, contextual understanding

APIs & Services

Google Places → reviews + locations
Google Directions → transit routing
Tavily / Firecrawl → social media mining
GrabFood / UberEats → food discovery
Supabase → personalization & storage

Example Workflow

Social Media → Tavily → Google Reviews → Reka Comparison → OpenAI Summary → User Recommendation

Challenges we ran into

Integrating multiple APIs into one coherent experience
Simulating food delivery recommendations without direct production API access
Translating raw transit data into simplified instructions
Balancing personalization while maintaining fast response times
Designing AI flows that feel contextual instead of generic

Accomplishments that we're proud of

Unified multiple travel use cases into one experience
Built multimodal interactions using text, voice, and image input
Added fact-checking to social media recommendations
Created travel guidance that prioritizes context, not just information
Delivered a working full-stack AI application

What we learned

AI becomes more useful when orchestration is done well, not just by adding more models
Context and personalization matter more than generating more content
Travel problems are often experience-design problems rather than data problems
Humanizing outputs can matter more than collecting additional information