Introducing: Postgres Best Practices
We are releasing Agent Skills for Postgres Best Practices to help AI coding agents write high quality, correct Postgres code.
Supabase Vector Buckets provide S3-backed storage for up to tens of millions of high-dimensional vectors per index with built-in similarity search using cosine, euclidean, or L2 metrics.
Users create buckets and indexes via dashboard or SDK, insert vectors with optional metadata, and query via Supabase clients or Postgres for k-NN searches and filtering. They complement pgvector for large-scale datasets in semantic search, recommendations, and RAG, targeting sub-second performance for most workflows.