H2O Wave and h2oGPTe: Build a YouTube Chatbot App
- Generative AI
- Enterprise h2oGPTe
This lesson walks through the code behind a YouTube chatbot built with H2O Wave and Enterprise h2oGPTe. It shows how the app takes a YouTube link, extracts the video's captions, and uploads that transcript into an h2oGPTe collection so a chatbot can answer questions grounded in the video's content. The lesson also covers Wave concepts used in the app, including zones for the header, content, and footer, streaming the LLM's response into the UI as it generates, and functions for adding and clearing cards.
Key takeaways
- How the app extracts a YouTube video's captions and uploads them into an h2oGPTe collection for chat
- How streaming updates the chatbot card with the LLM's response as it generates rather than all at once
- Why Wave zones like header, content, and footer control where each card appears in the layout
- What the addCard and clearCards helper functions do when building and resetting a Wave page
Start owning your AI