Langchain
Cheerio
Puppeteer
OpenAI
Supabase
TailwindCSS
Shadcn
Next.js
WatChat uses Supabase VectorDB retriever to find relevant information about the University of Waterloo and passes it as context to GPT-4o to provide an answer. To gather all the documents I used Puppeteer to scrape the pages in the University of Waterloo website, and Cheerio to clean up the HTML. I then used Langchain to split and chunk the data and store it in a Supabase VectorDB.
