The Document Center becomes a drive
Folders, drag-and-drop, rename, move, Recent and Shared. Bank statements get their own document type, check scans file against the feed on their own, and a statement parse that does not foot becomes a workbench instead of a dead end.
Zeno’s document center was built for one job: catch inbound paperwork, parse it, and file it against a record. It did that well and was useless for the other half of an organization’s documents. The lease that isn’t a transaction, the insurance certificate, the signed operating agreement, the folder of closing documents somebody needs to find in two years.
Those were living in Dropbox, and the bookkeeping was living here.
Folders, drag-and-drop, and a breadcrumb trail
The document center works the way a drive works. Folders nest inside the center itself, not only in the sidebar. Drag a file where it goes. Select several and move them. Rename them. Recent and Shared are in the rail. A toolbar and a breadcrumb trail replaced the wall of filters. Escape closes the menu.
Uploading a plain file is its own action now. A document you’re storing shouldn’t enter a parse queue and wait to be classified. You set its type at upload, from the types your organization files or one you define, and changing a label later doesn’t trigger a re-parse. Share and send are on the document itself.
Every file still has one destination and one canonical location in storage, so a document filed here and a document that arrived as a bill are the same kind of object with the same address. A document opens the record it belongs to, and a record’s attachments are the same files you see in the drive.
Bank statements get their own type
A bank statement is its own document type, filed under a canonical folder and distinguishable from a credit card statement, which parses differently and reconciles differently. Staging a statement and reconciling one are separate destinations, rather than one path with a flag on it.
Check scans file themselves
A scanned check matches itself against the bank feed and files against the transaction it belongs to, using the fields already read off it. Re-sending the same scan changes nothing, where it used to produce a second document.
When a statement doesn’t foot
The most common failure in statement import is arithmetic: the parse produces lines that don’t add up to the statement’s own totals, usually because a page was missed or a column was misread. That used to be a dead end, a red message and a suggestion to try again.
It’s a workbench now. When a parse doesn’t foot, Zeno does the arithmetic to name the culprit before anyone re-reads a PDF: which page, which side, and how much is missing. The AI proposes what it thinks is missing, and you decide. If you conclude the parse is right and the statement is odd, you can override, and the override is recorded as an auditable event someone can find later. An untyped card line fails the gate; Zeno will not pick its bucket by guess.
CSV import got a smaller version of the same treatment. The column mapping it guessed is shown to you before it’s acted on, and an empty mapping is reported as “nothing detected” rather than treated as a valid mapping of nothing.
Live for every organization. [email protected].