Point BeforeQuery at Google Drive and it indexes the documents you choose, Google Docs and Sheets included, with citations back to the original files.

​
Prerequisites

  • A Google account with access to the content you want indexed.
  • A BeforeQuery workspace owner role.

​
Data ingested

When you connect Google Drive, BeforeQuery indexes, for each file you select:
  • The file’s Drive link — what citations point at
  • Title, taken from the document’s own first heading and falling back to the file name
  • Content as markdown
Google-native files are exported first: Docs as markdown, Sheets as a workbook so every tab is indexed, Slides as text. Uploaded .pdf, .docx, .xlsx, .xlsm, .xltx, .xltm, .md, .mdx, .txt, .html, .htm, .csv, .tsv files are read in place. Anything else — images, video, Forms — is skipped and listed on the source so you can see what was left out.

​
Set up the connector

1

Add the source and connect

Open Knowledge Base → Sources → Add source, choose Google Drive, and click Connect Google Drive. Approve read-only access — nothing is written to your Drive.
2

Pick folders and files

The picker lists your Drive and searches it by name. Tick whole folders, individual files, or a mix.
A source with nothing picked indexes the entire drive. Picking your first folder therefore narrows it.
3

Add

Click Add — the first sync starts immediately. Folders are walked recursively.

​
Change what’s indexed later

A Drive source’s page has a Drive contents tab: what it indexes now, resolved to real names, with a picker to add more and a control to stop indexing one. You can also paste a link instead of browsing — a folder URL, or a link to a Doc, Sheet or Slide deck. Each is resolved against Drive as you add it, so an unreadable pick is refused now rather than silently indexing nothing later. Removing an item deletes nothing immediately: the next sync stops producing those documents and reconciliation removes them then, under the 45% review threshold.

​
What gets indexed

Google Docs, Sheets and Slides are exported and extracted, along with .pdf, .docx, .xlsx, .xlsm, .xltx, .xltm, .md, .mdx, .txt, .html, .htm, .csv, .tsv files. A Sheet is exported as a workbook, not CSV — a Sheets CSV export is the first tab only. Anything the pipeline can’t read is reported as a skipped item on the source.

​
Service accounts and shared drives

The service account fields are folded away under the connect button, for the case a consent screen for one user can’t reach: domain-wide delegation across a whole Google Workspace.

​
Keeping content in sync

Google Drive re-syncs every 10 minutes — there is no schedule to configure, and Sync on the source’s page runs one now if you cannot wait. Only changed content is re-embedded. See Sources Overview.

​
Troubleshooting

  • Sync succeeds but finds nothing? With a service account, the folder was never shared with its address, or impersonate user is unset. With the connect flow, check the Drive contents tab.
  • A pasted link was refused? It names a file type the pipeline can’t read, or one the connected account can’t open. The message says which.
  • More documents than expected? Nothing is picked, so it’s indexing the whole drive. Pick a folder to narrow it.
  • Sensitive files in the folder? PII masking applies at ingest, and you can mark documents restricted.