This FAQ chatbot is built on knowledge scraped from web pages or loaded from additional documents you provide. It accurately answers user queries and includes a caching system to reduce response time, helping limit the number of OpenAI API calls. Additionally, it features dynamic responses to keep users engaged while they wait for a reply. This chatbot also supports queries in almost any language, just like ChatGPT.
When ChatGPT started gaining attention, I was inspired to build a system that could automatically respond to users and support them on familiar websites and apps using GPT technology. During my research, I came across LangChain, which provided an API-based framework well-suited for this kind of application. This project was born from the idea of integrating intelligent, automated assistance into everyday digital experiences using modern language models.
- Open create_embeddings.py and update the URLs you want to scrape. By default, the URL points to the OpenPIM Wikipedia page.
- Run $ python create_embeddings.py.
- Run $ python create_cache.py.
- You can now integrate this chatbot into your web app or website to improve customer support and make it more interactive.
⚠️ You must use your own OpenAI API key.
Server: Python3.8 (Django)
Supported Python versions: 3.7–3.11 (due to dependency requirements)
Frontend: Plain JavaScript
Database: SQLite, Chroma database
used api: OpenAI