# Vector search in Postgres vs dedicated engine for RAG

> Vector search in Postgres vs dedicated engine can change cost, speed, and launch risk. Compare writes, ranking, and day-to-day upkeep.

## Why this choice gets expensive fast

The first RAG stack you ship usually stays in place longer than planned. Teams build scripts, indexing jobs, and content workflows around it, then product decisions start to depend on those habits. A choice that feels temporary in week one can shape the next six months.

The hidden cost often shows up in the write path. If new documents, edits, or deletes reach search late, retrieval goes stale fast. A user asks about a policy, a product spec, or a support note, and the system answers from last week's version. That looks like a model problem, but the real issue is often the path from content change to searchable vectors.

That is why