Xerum compared with enterprise search
Indexing everything is not the same as knowing anything
Enterprise search connects to all of it and ranks across all of it. That is genuinely useful for finding a document. It is a weaker foundation for an agent, because there is no version history to read at, no curation deciding what belongs, and no way for an agent to contribute back.
What enterprise search is genuinely good at
Coverage nobody else has
Dozens of connectors, years of work on each, and an index over systems you had forgotten you were paying for. If the question is 'where is that file', it is very hard to beat.
Permissions across many systems
Mirroring the access model of a dozen source systems into one index is difficult, unglamorous work, and the serious products do it properly.
It fits how companies already work
It asks nobody to change anything. Nothing has to be curated, approved or migrated — which is precisely why it gets deployed and why the results look like the drive it indexed.
The same question, in each
What is in the index?
Everything it can reach, including the stale, the superseded and the four near-identical drafts.
Resources recorded from a source, plus what an owner approved through a proposal. Curated by construction.
Can I read it as of last quarter?
No. The index reflects the current state of the documents it crawled.
Yes — reading at an older revision is an ordinary query, because nothing is edited in place.
Can an agent contribute back?
No. It is a read path over systems that own their own content.
Yes, through a proposal an owner decides on. The agent-write path is the product.
What does an answer cite?
The document it ranked. Which version it was is not part of the answer.
The revision, the resource and the origin URL — a specific state of a specific thing.
How are permissions applied?
Usually mirrored per source system into a shared index, then filtered at query time.
Pre-filtered inside the query, in an index that only ever held one team's data.
Where does the index live?
A shared, multitenant index with a customer dimension.
Your own container and index directory. No shared index exists to filter.
Keep your enterprise search
It answers a question Xerum does not: where, among everything this company has ever produced, is the thing I am looking for. That is a discovery problem and coverage wins it. Xerum answers a different one — what has this team established, when, and on whose authority — and that needs a curated record with versions and an approval loop, which is not something you can retrofit onto an index over a drive.
A record, not an index over everything you own
Built today, not on a roadmap. Every property above is already how the store works.