The Data Room
What ships today — the help centre — and what the wider standalone knowledge product is not yet.
The Data Room is Auralis' knowledge product: a place your content lives, with a taxonomy you own, that can feed an AI agent selectively rather than by default.
That is the direction. This page is about the part of it you can use, because the gap between the two matters more than the ambition does.
What ships today
One surface: the help centre, at Knowledge in the sidebar
(/help-centre). It is a public documentation site attached to your workspace,
with four screens behind it:
| Screen | What it is for |
|---|---|
| Overview | What this help centre holds, whether customers can reach it, and what has been imported into it. |
| Content | Every article, one locale at a time. Filter, sort, publish, and hand articles to the AI. |
| Collections | How articles are grouped and ordered for visitors. |
| Branding & domain | How the public site looks, and the address it is read at. |
A fifth row, Help Center, leaves the app and opens the live site. It is a plain label rather than a link when the workspace has no public address yet, because a guessed link is worse than an honest gap.
What is not built
Stated plainly, because "Data Room" promises more than the four screens above deliver:
- There is no standalone document repository. No general file store, no per-customer taxonomy beyond the help centre's own collection tree, no ingestion of anything that is not an article you wrote or imported.
- There is no separate knowledge-base surface. An earlier build had one, with its own ingest sources. It is gone; Knowledge in the sidebar now points at the help centre and nothing else.
- There is no visitor sign-in. The help centre has no end-user authentication, which is why an article the source system marked "signed-in only" is labelled Hidden: staff only here — it is served to nobody but your own staff.
Design work on the wider product is deliberately deferred. Nothing in these guides describes it, and you should not plan around it.
The AI feed is selective, and it is manual
This is the part people most often assume works the other way round.
Your help centre and your AI agent's knowledge base are two different stores. Publishing an article does not put it in the agent's index, and nothing puts it there automatically — not an import, not an edit, not a publish, not a visibility change. Handing an article to the AI is always an act a person performs, on a selection they made, with a confirmation that names the destination agent.
Why it is built this way
The credential used for the handoff does not check who owns the knowledge base it is told to write into. A wrong destination would publish your articles into another customer's assistant, and nothing downstream would catch it. So the destination and the exact count are on screen together before anything is sent.
See Handing articles to the AI for the mechanism, the two preconditions and the states an article moves through.
Who can do what
Writing, publishing and deleting articles is editor-level work. Collections management and branding are gated too, and where you are below the floor the controls are absent rather than greyed out — the screen shows the same answer the runtime would give rather than offering an action that comes back refused.
Next
Connecting Zendesk, Freshdesk and Intercom
The three credential connections: what each needs, how the test-before-store rule works, and testing and disconnecting.
Writing and publishing articles
The Content list and the article editor: modes, the toolbar, AI drafting, the details rail, share links, slugs and publishing.