You probably don't need another database.
Summary
The article argues that Postgres can cover caching, queues, search, documents, and vector embeddings, reducing the need for multiple data stores. It highlights Postgres features like unlogged tables, materialized views, jsonb, and pgvector as replacements for Redis, Elasticsearch, MongoDB, and Kafka, and cautions about the maintenance cost of multi-database stacks. It advocates evaluating Postgres before adopting specialized databases, especially for SMB-scale applications.