Comparisons
How Xerum compares
Agent memory, enterprise search, platform-native memory and plain RAG all answer a real question — just not the same one. Four comparisons, written category by category and naming no vendor.
Memory remembers what your agent said
Agent memoryA memory layer makes one agent better at continuing its own conversation. It is not a record of what your team knows, and it was never trying to be.
RAG is a technique, not a system of record
Your own RAGYou can build the retrieval in a fortnight. The store underneath it — versioned, curated, permission-aware — is the part that takes a year.
Search over the swamp is still the swamp
Enterprise searchIndexing everything is a different bet from curating something. One gets you coverage; the other gets you an answer you can act on.
One vendor’s assistant, one vendor’s memory
Platform memoryThe platforms validated the category and then built it per-assistant. Cross-agent and team-level is the part they did not build.
Looking for how it fits your own field? Who it's for →