All resources
Data & RAG20 August 2026·5 min read

Keeping a RAG knowledge base fresh: ownership, versions and expiry

Stale content is the quietest cause of wrong answers. A maintenance routine that keeps the index aligned with reality.

A RAG system is only as current as its sources. Policies change, products are retired and old versions linger in the index, so the system answers with outdated confidence.

Routine

  • Owner per content area, responsible for accuracy.
  • Version and effective dates stored as metadata, with superseded documents removed or flagged.
  • Automatic re-indexing when a source changes.
  • Review dates — content past its review date is flagged or excluded.
  • Feedback loop — 'this answer is wrong' reports are routed to the owner.

Show the source and date in every answer. Users then spot staleness themselves and trust the system more.

Related capability

This is what our AI Data Strategy & ML practice is built around.

See how it's scoped

Keep reading

Data & RAG

RAG, fine-tuning or a long context window? Choosing by the problem

6 min readData & RAG

Permissions-aware retrieval: making sure the assistant only shows what each person may see

6 min readData & RAG

Reranking: when a second pass is worth the extra latency

5 min read