🌍 Tambo Earth

Tambo Earth is an interactive, AI-powered 3D globe application that allows users to explore the world through natural language conversations. It combines the visual immersion of 3D globe with the intelligence of a generative AI assistant, creating a seamless educational and exploratory experience.

✨ About the Project

The goal of Tambo Earth is to reimagine how we interact with geographical data. Instead of static maps or search bars, users can ask questions like "Show me a culinary tour of Italy" or "Where should I go for a tropical vacation?". The AI not only answers but controls the globe to fly you there, displaying rich, interactive content.

Key Features

• Interactive 3D Globe: Powered by react-globe.gl, featuring a realistic day/night cycle, atmosphere, and interactive country polygons.
• AI Pathfinder: Integrated with @tambo-ai/react, the AI assistant acts as your guide, generating custom UI components on the fly.
• Generative UI: The chat interface renders interactive components like:
  • LocationCard: Details and photos of visited places.
  • FunFact: Interesting trivia about locations.
  • JourneyList: Curated travel itineraries with clickable destinations.
• Seamless Navigation:
  • Click-to-Fly: Click any location in the chat to instantly fly the globe there.
  • Smart Zoom: Interactive zoom levels for selection (1.5) and visiting (0.5).
  • Label Interaction: Persistent country labels act as "back buttons" to zoom out.
• Gamification: Track your "Visited Countries" and unlock ranks from Wanderer to World Class.

🛠️ Tech Stack

• Frontend: React, TypeScript, Vite
• Styling: Tailwind CSS
• 3D Visualization: react-globe.gl (Three.js wrapper)
• AI & GenUI: @tambo-ai/react, @modelcontextprotocol/sdk
• Markdown Rendering: react-markdown, remark-gfm
• Icons: lucide-react
• Validation: zod

🏗️ Architecture

The application is built around a centralized context and event-driven architecture to sync the 3D view with the AI chat.

• Context (CountryContext): Manages global state such as:
  • visitedCountries: List of visited locations for gamification.
  • targetCountry: A signal used to trigger globe navigation from arbitrary components (like chat).
  • flyToCountry(): Function exposed to components to control the globe camera.
• Components:
  • GlobeView.tsx: Handles the 3D rendering, camera controls, and polygon interactions. It listens to targetCountry changes to animate flight.
  • ChatSidebar.tsx: Manages the AI thread. It renders Markdown and custom "Tools" (Generative UI components) called by the AI.
  • App.tsx: The main layout that orchestrates the tool registration and initial prompt handling.

🚀 Getting Started

1. Clone the repository:

   git clone https://github.com/your-username/tambo-earth.git
   cd tambo-earth
   

2. Install dependencies:

   npm install
   # or
   pnpm install
   

3. Set up Environment: Create a .env file and add your Tambo API key:

   VITE_TAMBO_API_KEY=your_token
   

4. Run Development Server:

   npm run dev
   

📈 Learning & Growth

Developing Tambo Earth presented several unique challenges:

• Streaming Stability: Handling streaming AI responses with complex tool calls required robust schema definitions using zod. We moved from loose z.any() types to strict z.string().nullish() to ensure reliability.
• 3D-2D Sync: Synchronizing React state (2D UI) with the imperative Three.js camera (3D Globe) required careful use of refs and useEffect hooks to bridge the gap.
• UX Refinement: We iterated significantly on the "Visit" flow, moving from simple clicks to a sophisticated "Select -> Visit -> Persist Label" model that feels natural and keeps the view uncluttered.