Skip to content

About

Simple Agentic LLM assistant for planning trip with Overpass API

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

WandeRound

AI-powered trip planning assistant with interactive maps. Ask in natural language — WandeRound finds places, builds a route, and shows it on a map.

Stack: FastAPI · LangGraph · DeepSeek · Next.js · Tailwind CSS · OpenStreetMap

Wanderound Demo Video

Watch the Demo Video


Local Development

You need two terminals running at the same time.

Prerequisites

1. Clone & configure

git clone https://github.com/nii0708/WandeRound.git
cd WandeRound

Copy the env template and fill in your API key:

cp .env.example .env

Edit .env:

DEEPSEEK_API_KEY = sk-your-key-here
DEEPSEEK_MODEL   = deepseek-v4-flash

2. Start the backend (Terminal 1)

pip install -r requirements.txt
python server.py

Backend runs at http://localhost:8000

3. Start the frontend (Terminal 2)

cd frontend
npm install
npm run dev

Frontend runs at http://localhost:3000

Open http://localhost:3000 in your browser.


Deployment

Backend → Render

  1. Go to render.com and sign in

  2. Click New → Web Service

  3. Connect your GitHub repository

  4. Fill in the settings:

    Field Value
    Root Directory (leave empty)
    Runtime Python 3
    Build Command pip install -r requirements.txt
    Start Command python server.py
  5. Go to the Environment tab and add:

    DEEPSEEK_API_KEY = sk-your-key-here
    DEEPSEEK_MODEL   = deepseek-v4-flash
    
  6. Click Deploy

  7. Wait for the build to finish, then copy the service URL — it looks like https://wanderound.onrender.com

Note: The free tier spins down after 15 minutes of inactivity. The first request after idle takes ~30–60 seconds to wake up. This is normal.


Frontend → Vercel

  1. Go to vercel.com and sign in

  2. Click Add New → Project

  3. Import the same GitHub repository

  4. Important: change the Root Directory to frontend

  5. Framework will be auto-detected as Next.js

  6. Go to Environment Variables and add:

    NEXT_PUBLIC_BACKEND_URL = https://wanderound.onrender.com
    

    Replace the URL with the one you got from Render.

  7. Click Deploy

Your app is now live at the Vercel URL.

Every push to main automatically redeploys both Render and Vercel — no manual steps needed.


Environment Variables Reference

Variable Where Description
DEEPSEEK_API_KEY Backend (Render) API key from platform.deepseek.com
DEEPSEEK_MODEL Backend (Render) Model name — use deepseek-v4-flash
NEXT_PUBLIC_BACKEND_URL Frontend (Vercel) Your Render backend URL

Project Structure

WandeRound/
├── app.py              # LangGraph agent (trip planning logic)
├── server.py           # FastAPI backend with SSE streaming
├── thread_manager.py   # Chat thread persistence
├── tools/
│   ├── osm.py          # OpenStreetMap / Overpass API
│   └── route.py        # Route optimization (TSP)
├── data/               # GeoPackage files (auto-created, gitignored)
├── frontend/           # Next.js frontend
│   └── src/
│       ├── app/        # Pages and layout
│       ├── components/ # ChatSidebar, ChatWindow, MapView
│       ├── lib/        # Utilities
│       └── types/      # TypeScript types
└── requirements.txt    # Python dependencies

Credits

I'm really thankful to the OSM, Overpass, and Nominatim communities so that this project can happen. Thx to @AshecOne that tested this code and ISHteam that helped me a lot in 2025.

Built with OpenStreetMap · Overpass API · DeepSeek · LangGraph

About

Simple Agentic LLM assistant for planning trip with Overpass API

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages