Discover
Let the agent surface what's interesting in your data.
Discover is a quick way to see what's already available to you in a warehouse — the documents you've uploaded, the data files your pipelines have landed, and the tables in your datalake — and pull any of it into a Databasin One conversation.
Where it lives
Discover is a panel in the Databasin One sidebar. Open it while you're chatting and it sits alongside your data context, so anything you find can go straight into scope for your next question.
What it needs first
Discover reads from a warehouse, so two things have to be in place:
- A selected data warehouse — if you haven't picked one, Discover tells you to.
- A running cluster — Discover won't wake cold compute on its own, so you start the cluster first, then discover.
Once both are set, pick a catalog and choose Discover. It scans that catalog and organizes what it finds into a browsable folder tree.
Docs, Files, and Datalake
Discover groups results into three modes you can toggle between:
| Mode | What it shows |
|---|---|
| Docs | Documents you've uploaded that have been embedded for search. |
| Files | Ingested data files that aren't embedded — spreadsheets, exports, and the like. |
| Datalake | A search across the catalog's tables and columns by name. |
The Docs and Files modes show a folder tree you can expand and browse; Datalake mode is a search box for finding a table or column when you already know roughly what you're after.
From a find to a conversation
Browse or search, select the documents or files you want, and add them to your data context. From that point they're in scope — you can ask the agent about them like any other table.
Everything you pull in is still queried under your own access. Discover helps you find data; it doesn't grant you any you didn't already have.