Browsing
Documents live in a two-pane explorer — a folder tree on the left, files on the right — scoped to your active project. Search by name, or filter by type, project stage, source, folder, and date added. Each row shows the document’s type, category, processing state, source, and current version.Uploading
Click Upload files or just drag files anywhere onto the page. When uploading, choose the document’s visibility:- Everyone in the organization
- This project
- Only me
What happens to every document
Each file runs through the same pipeline, whether uploaded or synced:- Stored and versioned — content-addressed storage; re-ingesting the same file never duplicates it, and each change becomes an immutable version.
- Parsed — PDFs (with OCR for scans), Word, Excel, and email formats, preserving page, line, and position.
- Classified — into construction document types: drawings, specifications, submittals, RFIs, contracts, change orders, invoices/pay applications, schedules, meeting minutes, daily logs, safety docs, permits, closeout docs, and more.
- Extracted — key fields pulled from contracts, drawings, submittals, invoices, RFIs, and change orders, each anchored to the exact spot in the file it came from.
- Indexed — chunked and embedded so the assistant can find and cite it, down to the page and line.
The document drawer
Click any document to open its details:- Extracted fields — each field links back to the exact page and position it was read from; low-confidence extractions carry a Review chip.
- Versions — the full version chain with an upload dropzone for new revisions, plus changes between versions summarized with materiality and affected pages.
- Supersede — mark a document as superseded by a newer one; superseded files are badged everywhere they appear.
- Provenance — who ingested it, when, from where, with content hashes and parse details.
- Activity — the audit trail of every action taken on the document.

