Yasnora AIRU
AI and knowledge
AI and knowledge · 2026.10

RAG and Memory

Retrieval, citations, scope, retention and memory policy.

2 min read
Reviewed 2026-10-25
Bilingual article
01

Retrieval and citations

Knowledge connects answers to authorized source documents. RAG retrieves relevant material within the current workspace scope and uses it as context for an answer.

If a source is missing, check its processing state, workspace binding and your role. A citation supports a specific passage, not every statement in the response.

02

Memory scope

Memory stores selected preferences or context when policy allows it. It can have personal, Group or organization scope and differs from a source-backed Knowledge store.

A missing memory control can reflect workspace policy. Personal context must not be treated as shared team knowledge without an authorized sharing path.

03

Source lifecycle

Files move through upload, validation and processing before becoming ready or failed. Uploading a file does not automatically make it available in every Knowledge store.

Use only files you are authorized to process. Index cleanup and retention may continue after a deletion request; do not assume immediate erasure from backups.

04

Entities and relationships

The Knowledge Graph presents entities and relationships with source provenance. It supports discovery but does not make an inferred relationship a verified fact.

Graph refresh or deletion follows the configured source lifecycle. A visually prominent relationship can still be an inference that needs verification.

Maintained by knowledge · Edited 2026-10-25 · Documentation 2026.10