UnisonAgent Memory vs RAG: What's the Difference?
RAG retrieves documents you indexed; agent memory accumulates what your agents learned while working. Different questions - most teams need both.
Most developers mistake RAG for agent memory - and using RAG for memory is exactly why agents keep forgetting your architecture decisions.
RAG retrieves from a corpus you indexed; agent memory accumulates what your agents and teammates learned while working. They answer different questions.
| RAG | Agent memory / team brain | |
|---|---|---|
| Source of truth | documents you already had | facts produced during work |
| Write path | batch indexing | continuous, from live sessions |
| Shape | chunks + embeddings | entities, facts, documents, links |
| Question it answers | "what do our docs say?" | "what did we decide, learn, and try?" |
| Staleness | re-index to update | facts supersede each other over time |
RAG cannot tell you that yesterday's session chose Postgres over SQLite and why - nobody wrote a document about it. A team brain can, because the deciding agent recorded the fact at decision time.
Where Unison sits
Unison is the memory side: agents ingest conversations, the pipeline extracts entities and bitemporal facts, and any agent recalls the relevant slice as markdown. If you already run RAG over your docs, keep it - the brain holds what work produces, your index holds what you wrote down beforehand.
Memory types
Episodic, semantic, and procedural memory for AI agents - what each type stores, when to use which, and how Unison classifies them at search time.
How Unison works
Architecture in one page: agents ingest conversations and documents, an async pipeline extracts entities and bitemporal facts into a knowledge graph, and any agent recalls the relevant slice as markdown.