🚀 Live Demo: AlecBytes/chatbot
📝 Currently building a chatbot that uses retrieval-augmented generation (RAG) to reason and respond with information outside of the model's training data. I kicked things off with a Vercel AI SDK template, tweaked the model prompt, and chunked & embedded my resume info in a vector database. After testing prompt responses on a working prototype, I'm now fine-tuning and adding more information chunks. Next up, I'll re-embed the updated chunks and see how they boost the chatbot’s responses.
If you're into AI or chatbots, let's connect and share insights!
If you find this project useful, consider starring ⭐ the repo! Your support helps others discover it and motivates further development. Thank you! 🙏
- Information retrieval and addition through tool calls using the
streamTextfunction - Real-time streaming of model responses to the frontend using the
useChathook - Vector embedding storage with DrizzleORM and PostgreSQL
- Animated UI with Framer Motion
- Next.js - React framework
- Tailwind CSS - CSS framework
- Radix UI - Unstyled, accessible components
- Framer Motion - Animation library
- Vercel AI SDK - Core AI functionality
useChathook for frontend chat interfacestreamTextfunction for streaming responses
- OpenAI API (GPT-4o model)
- PostgreSQL - Vector database storage
- DrizzleORM - Database ORM
This project is not open-source, and no license is granted for reuse, modification, or distribution.
If you are looking to build a similar chatbot, please refer to the official Vercel Guide and its starter code. The modifications in this repository are tailored to my specific needs, and you should create your own custom implementation.
The starter code and images used in this project are sourced from Vercel's Guide.