Platform · Drive
Everything you feed itstays usable.
Load the bundle or link Google Drive. Papers are indexed with their provenance intact, scans are read by OCR, and your own file is cited the way an Act is.
Drive is the context layer. A document that lands here is extracted, summarised, chunked, and indexed, and every retrieval that uses it can name the file and the page it came from. Folders, sharing, starred items, and a bin work the way you expect. What is different is that the contents are addressable by the agent, so a clause buried on page 41 of a scanned bundle is as reachable as a section of the statute it argues about.
Provenance travels with the passage, not just the file
From a file on your desktop to an answer you can check
Put the documents in
Drag a folder in, or link Google Drive and choose which files to index. Nothing is indexed that you did not select.Text is recovered, including from scans
Native text is extracted directly. Scanned and image-based pages fall back to OCR, so a photographed judgment is searchable like any other file.Each file is summarised and indexed
You get a short summary for triage and a chunked index for retrieval, so the agent can quote a passage rather than paraphrase a whole document.The agent retrieves with the source attached
When a run uses one of your documents, the answer names the file and the page. You open it and check, the same way you would check a citation.
What Drive handles
- Folders, starring, sharing, and a bin, with per-matter and per-workspace scoping
- PDF, DOCX, XLSX, TXT, and image files
- OCR fallback for scanned and image-based pages
- Per-file summaries for triage before you read anything in full
- Chunked indexing, so retrieval quotes a passage rather than a whole document
- Google Drive linking, with explicit per-file selection and a purge that removes what was indexed
- Retrieval that names the file and page behind every passage
- Storage usage visible per workspace
Questions
About Drive
- What happens to a scanned PDF?
- If a page has no extractable text, 8rney runs OCR on it before indexing. The recovered text is searchable and quotable like any other page, and the answer still points back at the page it came from.
- Does linking Google Drive index everything I own?
- No. You pick the files. Only what you select is fetched and indexed, and disconnecting lets you purge what was indexed from that account.
- Can the agent answer from my documents alone?
- Yes. A run can be scoped to the matter record, so the answer comes from your files rather than from public sources or the model's training data.
- Who can see a document I upload?
- Access follows the matter and the workspace roles it is in, ethical walls included. In an institution workspace, new records are private to the people explicitly granted access.
- Is Drive the same as the Library?
- Yes. Drive is the current name for the same surface. Old links to /library still resolve here.