Semantic Search with OpenSearch & Embeddings in NestJS (Step by Step) – Ep 11

How does a store understand that warm coffee on my commute means a thermal mug? With embeddings, and this episode builds that engine.

We add OpenSearch to docker compose, build an EmbeddingService on text-embedding-3-small, create the knn_vector index with the right mapping, and bulk-embed and index the whole catalog from the seed script. The concept of vectors as meaning is explained properly, because every AI episode after this one stands on it.