# MemoryOps vs RAG

> RAG retrieves information for a model. MemoryOps operates the durable memory layer around agent work.

MemoryOps asks what should be remembered, where it came from, who can use it, whether retrieval is correct, and how stale or unsafe context gets repaired.

## Different jobs

RAG is useful when an agent needs outside information from documents or databases. It answers, "what text should the model see for this query?"

MemoryOps answers a different question: "is the organisation's remembered context healthy, governed, and safe enough for agents to rely on?"

## Why RAG alone is not enough

A RAG system may retrieve the right document and still leave the agent guessing which decision is current, who approved it, whether it applies to this project, and whether the user is allowed to see it.

MemoryOps narrows the durable takeaway and keeps source, scope, redaction, conflict state, and audit attached.

## How Ambience fits

Ambience is the governed company-context layer around agent memory. It can sit beside search and RAG systems while preserving the memories agents should reuse directly.

For teams, the practical pattern is: use RAG for broad source retrieval, use Ambience MemoryOps for durable decisions, conventions, failures, skills, and references.

## Related

- [Ambience vs internal RAG](https://ambience.sh/compare/internal-rag)
- [MemoryOps](https://ambience.sh/company-context/memoryops)
- [ContextOps](https://ambience.sh/company-context/contextops)
- [Company context vs knowledge base](https://ambience.sh/company-context/company-context-vs-knowledge-base)