Org memory is a category that emerged when LLMs got good enough to synthesise across thousands of artefacts in seconds. It refers to the system that captures every piece of work an organisation produces (Slack threads, GitHub PRs, Linear tickets, Notion docs, customer support transcripts, meeting recordings) and makes them queryable as a single coherent body of knowledge.
What it isn't
Org memory is not a wiki. A wiki encodes one viewpoint at one point in time; org memory captures every artefact and lets the answer materialise from them.
Org memory is not enterprise search. Search returns documents and asks the user to reconstruct the answer. Org memory does the reconstruction and returns a sentence with citations.
What makes it real
Four properties separate a real org-memory system from a glorified RAG pipeline: passive capture (no behaviour change required), automatic connection (a graph of relationships, not a flat index), answerable retrieval (questions in, sentences out), and surviving citations (every claim points back at its source).
All four are non-negotiable. RAG without the connective tissue is a parlor trick.