Conferences often publish their sessions on YouTube after the event but finding what you're looking for in them can be time consuming.

With hours upon hours of video content how do you find that thing your colleague mentioned, that insight you wanted to revisit, or find something of interest from your topics?

Well, let's build this. Let's leverage some aspects of the current generation of AI models to create an app which doesn't create answers "from thin air" but contextualised around YouTube playlists from DDD, or really any data that you could want.

We'll look at how to extract intents from prompts and semantically search highly unstructured data to produce (hopefully) meaningful outcomes.