I've been working with embeddings recently.

Given some text, an embedding model returns a point in vector space.

First uses of embeddings were related to Topic clustering. What is the distance between to pieces of text in semantic space?

I came across the concept of Semantic Axes, where multiple points in vector space can be defined by two different sets of text - hot and cold, good and bad, service-related inquiry or website complaint.

Being able to create contextual framings is an interesting capability.