Cases in depth
Files route by identifier. A matching identifier joins the existing case, while a new identifier creates a new case. The app never overwrites your manual edits, and every case carries an activity timeline.
The app groups your incoming paperwork into cases. When you send a file, the app reads its identifier. This routes the file to the correct case.
How files route to cases
Every case in the workspace has a Case ID. When a new file arrives, the app checks this identifier. A matching identifier always joins the existing case. A completely new identifier creates a brand new case.
Different ingestion channels read these identifiers in different ways. Google Drive reads only the folder delivery metadata. Gmail reads the email, and falls back to file extraction. Manual uploads extract the identifier directly from the file.
When a file has no identifier, the app creates a new case. The case receives an auto-generated ID and an auto tag. If a later extraction finds a valid identifier, the case auto-promotes to that Case ID. If another case already claims that identifier, the collision leaves the auto ID in place.
If the app finds an existing case, it attaches the file. An automation can also move the matched case to a mailbox. This action creates a mailbox-move event in the activity timeline.
How extraction updates data
Extraction runs immediately on the new incoming file. The latest extraction replaces the previous value for a field. This includes empty or null values, which overwrite existing text.
Your manual edits to the case data are permanent. The app never overwrites a field you changed by hand. Date values normalize to the US MM/DD/YYYY format. Unrecognized free text stays unchanged in the field.
For repeated rows, Batch fields replace the entire array. They do not concatenate new rows into the existing list.
To run extraction again, open the Autofill this file panel in the review modal. Click Review actions > Re-extract to update the data.
When files route to the wrong case
A wrong read creates a new case instead of joining the correct one. The activity timeline shows Read as [ID] or Read from [source] as [ID].
Ask the AI Assistant to check a mismatched file. The assistant can re-extract a document ID or replay a Gmail subject and sender. It cannot replay an email body or Drive delivery metadata. The assistant does not move anything until you apply the result.
When you apply a re-read ID that nobody owns, the app renames a one-file case. It splits a multi-file case into a brand new case.
A chat instruction applies only to that specific file. Put general guidance in the AI instructions on the template details. This ensures future documents route correctly when they arrive.
When a move leaves an empty routing artifact behind, the app deletes it. The app only deletes the case if it has absolutely nothing. This means no files, notes, teammates, generated documents, or manual edits.
Moving and merging files
When an orphan case holds a document with a claimed ID, a merge banner appears. Click Merge into {case identifier} to move the files and delete the orphan. You can also click Dismiss to ignore the suggestion. A banner merge re-attaches source files through the target extraction rules. If the orphan holds a document your category access hides, the merge is refused with Some documents could not be moved; source case kept. and both cases stay as they are. Ask a teammate who can see every document on the orphan to merge it.
You can move a single file directly from the case. Open the collapsed Documents console to view your options. The console offers Upload, Fill from template, and Generate from template. Click Move to another case... on the stray file.
Use Search cases by ID to find a destination. You can search for a full ID outside the recently loaded list. A moved file keeps its extracted values in the new location.
You can also click Start a new case from this file. The dialog prefills the read ID for your convenience. Leaving the field blank assigns an ID automatically. If you type an ID that already exists, the app asks to move the file there.
Moving a file, starting a new case from it, correcting its ID, changing its category, and deleting it all require that you can see the file's category. A file hidden from you cannot be acted on.
To fix a read error in the timeline, click Correct file ID. The app offers Move file there or Move to new case {identifier}. A correction matching the current or historical Case ID is a provenance-only update. This specific correction updates the record but does not move the file.
Renaming and uploading manually
To rename a case, click Case actions > Edit case. Change the Case ID and click Save changes. If the new name conflicts, the app offers Merge into {case identifier} or renaming to a variant.
To add a file manually, open Edit case and find the Upload File section. Choose a Source File in PDF or image format. You can type an optional File Description for the document. Click Upload to Case to finish the manual addition.
If the case already shows a file with the same name, you choose Save as New or Overwrite. Save as New stores the upload with a number added, like scan (1).pdf. The app also adds a number when the name belonged to a file that was deleted from the case, so an upload never replaces a deleted or moved file. Overwrite replaces the existing file only when you can see every file stored under that name; otherwise the upload is saved with a number added.
The activity timeline
The Activity timeline shows events with the newest first. The app derives this timeline from multiple different sources.
Each row is a short sentence under the person or source that acted. A file row says how the file arrived: received, generated, or uploaded. Its category chip sits beside the file name. A field edit names the field and, for teammates, the value it replaced and the new value, for example "Changed Due date from Mar 3 to Mar 10". Edits made through the API or by an automation appear under that source's name.
Teammates and clients use a permission-gated composer to add notes. Notes contain text, optional event dates, and file attachments. Notes keep an edit history, and you can delete them.
Dates in the timeline offer View in calendar and Download .ics. Signed documents show an electronic-signature count on the timeline. They preserve the signed PDF and an immutable consent and hash record.
If a teammate lacks access to a file category, the app hides the related activity. The case itself remains visible to the restricted teammate.
Everything except notes is a permanent timeline record. The app offers no delete control for case creation, template runs, or file ingestions.