Skip to main content

​
Work With Documents in a Vault

Each Document Store opens to a document table. The table combines the original files, processing status, and AI-extracted metadata columns you configured for the store. This is where you upload new files, monitor processing, search across content, review extracted fields, preview documents, and export or download results.

​
Adding Documents

You can add documents in three ways:

Upload manually

Click Add Documents and choose files from your computer.

Drag and drop

Drop supported files directly into the document table.

Sync folders

Import documents automatically from connected Google Drive or OneDrive folders.
Supported upload types include common document, Office, spreadsheet, presentation, text, CSV, HTML, Markdown, RTF, and EPUB files. Manual uploads are limited to files up to 25MB. When files are uploaded, they are queued and processed. The table shows processing states while metadata extraction runs.

​
Document Table

The document table is the main working surface for a store.

​
Search and Retrieval

Document search uses hybrid retrieval across document text and metadata. This means you can search for exact business fields, document names, and semantic concepts from the file content. Examples:
  • “leases expiring next quarter”
  • “Acme vendor invoices”
  • “documents mentioning termination for convenience”
  • “policies owned by security”
  • “meeting notes with budget action items”
If no results match, clear the search to return to the full table.

​
Previewing Documents

Click a document to open it in a document preview. The table remains available, and open previews appear as closable tabs. Preview tabs are useful when you need to check the source document while comparing extracted metadata in the table.

​
Bulk Actions

Select one or more rows to take action on documents.
Deleting documents removes them from the store. Export metadata or download files before deleting anything you need to preserve.

​
Processing Status

Uploaded and synced files are processed before their metadata is fully available. While processing, documents may show queued or in-progress states. Failed documents can be identified in the table and retried by re-uploading or re-syncing after resolving the source issue.

​
Best Practices

  • Upload related documents into focused stores instead of mixing unrelated material.
  • Check processing status after large uploads or sync imports.
  • Use search for semantic discovery and filters for structured review.
  • Export CSV metadata when you need reporting or spreadsheet analysis.
  • Preview source files when reviewing important AI-extracted fields.
  • Delete carefully; use bulk actions only after confirming the selected rows.