Journeo

---

About

This is the backend for Journeo, an advanced LLM-based travel itinerary planner. It leverages OpenAI's GPT models through LangGraph, Google Places API, and Mapbox GL JS to generate personalized, intelligent travel itineraries with precise geolocation details.

Key Features:

• Intelligent Itinerary Generation: Uses OpenAI's GPT to create personalized travel plans
• Preference Extraction: Advanced agent-based system to understand and extract user travel preferences
• Geocoding Services: Integrates Google's Geocoding and Places APIs to get precise location details
• Route Calculation: Uses Mapbox Directions API to calculate routes between destinations

Complementary frontend can be found in the Journeo Frontend Repository.

---

Technologies Used

• Python
• FastAPI
• LangGraph
• OpenAI GPT-4o
• Google Places API
• Mapbox Geocoding API

---

Setup / Installation

Note: This is the backend component of Journeo. You must also set up the Journeo Frontend for a complete application experience.

1. Prerequisites:

   • Python 3.8+
   • OpenAI Account
   • Google Cloud Platform Account
   • Mapbox Account

2. Obtain necessary API keys:

   • Mapbox access token
   • OpenAI API Key
   • Google Cloud API Key with Places API enabled

3. Clone the repository:

   git clone https://github.com/eugene-lok/journeo-backend.git
   cd journeo-backend

4. Create a virtual environment:

   python3 -m venv env
   
   # Activate the environment
   # On Linux/Mac:
   source env/bin/activate
   
   # On Windows:
   \env\Scripts\activate

5. Install dependencies:

   pip install -r requirements.txt

6. Set up environment variables:

   • Create a .env file in the root of your project
   • Add the following keys:

   MAPBOX_ACCESS_TOKEN=your_mapbox_access_token
   OPENAI_API_KEY=your_openai_api_key
   GOOGLE_API_KEY=your_google_api_key

7. Run the FastAPI server:

   uvicorn app.main:app --reload

8. The API will be accessible on http://localhost:8000

Frontend Repository

For the client-side implementation, visit the Journeo Frontend Repository.

Key Components

Preference Extraction Agent

• Uses LangGraph to create a state-based workflow for extracting travel preferences
• Handles complex, multi-turn conversations to gather trip details
• Ensures comprehensive information collection

Itinerary Generation

• Leverages OpenAI's GPT-4o for intelligent itinerary creation
• Generates day-by-day travel plans with precise location details
• Calculates budget breakdowns and provides comprehensive travel guidance

Geocoding and Route Services

• Uses Google Places API to get detailed location information
• Retrieves coordinates, place details, and photos
• Calculates routes between destinations using Mapbox Directions API