Embeddings / Vector Search
Problem
A user searches:
“How do I stop services from overwhelming each other?”
The document says:
“Apply backpressure.”
The words don't match, but the meanings do.
Solution — Embeddings
Represent semantic meaning as vectors and search for nearby vectors.
query meaning
↓
vector
↓ similarity
document vectors
Core idea: vector search retrieves information by semantic similarity rather than only matching words.