Similar Shows Embeddings

[wesbos]

Each show will have its transcript run through Open AI embeddings API, and that returned vector is stored in a DB.

Then with each vector we can do two things:

1. recommend similar shows by searching for nearest neighbor
2. We can even make a chat bot that answers questions like "What does Scott think about CSS variables?".

[wesbos]

cloudflare workers now lets you store embeddings in their DB. So this might be a nice lightweight solution to both generate the embeddings + query them.

I will test to see if openai embeddings are better though