🧠 Ask Android Blog – RAG Agent

A lightweight Retrieval-Augmented Generation (RAG) application built with LangChain, designed to answer questions based on the official Android Developers Blog. This project comes preloaded with content from 50 blog pages, providing ample context to power intelligent, blog-aware responses.

🚀 Getting Started

1. Clone the repository

git clone https://github.com/amsavarthan/ask-android-blog-rag-agent

cd ask-android-blog-rag-agent

2. Add your API key

Create a .env file and include your Google API key (required for Gemini or other LLM providers):

DOCKER_HOST_PORT=
GOOGLE_API_KEY=

3. Start the app with Docker

Ensure you have Docker and Docker Compose installed. Then run:

docker-compose up -d

Visit the app at http://localhost:8000

💡 Note The app comes with 50 pages of pre-scraped blog content. You can ask questions related to the topics covered in those posts.

🛠️ Customization

🔄 Switching to Ollama or Other LLMs

To use Ollama or another LLM instead of Gemini, open config.py and modify the get_llm() function accordingly. You can also change the embedding model from the same file.

♻️ Enable Blog Refresh in the UI

To allow dynamic blog updates from the sidebar, set ALLOW_REFRESH = True in config.py.

⚠️ Important: Use caution when setting a high number of pages to scrape. Excessive requests may put unnecessary load on the Android Developers Blog.

📂 Project Structure

├── src/                  
├───── app.py               # Streamlit frontend
├───── config.py            # App Configuration
├── context/                # Preloaded vector store
├── .env                    # Environment variables
├── docker-compose.yml      # Container config
└── requirements.txt        # Python dependencies

📬 Feedback

If you encounter bugs or have feature suggestions, feel free to open an issue or submit a pull request.

📖 License

This project is open source under the MIT License.