BUILD A SEARCH ENGINE, NOT A VECTOR DB

Summary from the TLDR Web Dev newsletter

Vector database startups have proliferated in the last 12 months. However, vector databases should not be solely seen as a way to expand a language model's memory, but rather should be thought of as a specialized search engine. While embeddings and vector search can enhance search results, building a robust search engine with a combination of high-quality embeddings and traditional search methods is a better approach to utilizing AI well.