Where Invoices Live: The Vault

Invoices moved. Here is where they actually live now.

If you have a bookmark for /invoices, it still works, it just does not stop where it used to. Open it and Verinode redirects you straight through to /documents, which itself redirects to /data. That second hop is not a bug either: Documents was renamed to Data (the sidebar calls it Vault), and that is the real, permanent home. Two redirects, one destination: everything Verinode has parsed out of every document you have sent it, invoices included.

The reason invoices do not get their own page is structural, not cosmetic. An invoice is one of many document types the ingestion pipeline reads, alongside certificates, estimates, renewal notices, statements, payment notices, job exports, carrier scorecards, and incident reports. All of them go through the same parser, land in the same table, and feed the same signals. Giving invoices a standalone screen would mean showing you a slice of the Vault under a different name. Instead, the Vault shows you the whole picture, and you filter down to invoices whenever that is the view you want.

Where to find it

Open Vault from the sidebar. In the browser it lands on /data (the old /documents and /invoices paths both forward here automatically).

The Vault is one of the sections built on Verinode's shared page shell: a hero panel at the top, a row of tiles you click to drill in, and a row of your most recent captures. Clicking a tile does not navigate anywhere, it opens an overlay (Verinode calls this a slider) with four tabs across the top: Uploads, Tracking, Lists, Notifications. Invoices live inside the Uploads tab, and the Tracking tab shows what Verinode has learned about the vendor, client, or job each invoice is attached to.

The hero panel: "What Verinode Is Tracking"

The first thing you see is a large number, the total count of documents Verinode has captured for your business, counting up on load. Beside it, once you have at least one document, a colored pill reads something like "82% Extraction Completeness": the share of extractable fields Verinode actually pulled out across your documents, not a confidence score, a completeness score. That pill turns green at 80% and above, amber from 40% to 79%, and red below that.

Under the number, one line of context changes based on your state:

• With zero documents: "Connect a channel, email, uploads, voice, or photo, and your vault starts filling."
• With documents that need a quick assignment: a note telling you how many need review to unlock more signals.
• With documents still processing: a note telling you how many are still pending, with signals arriving as soon as they land.
• Otherwise: "Feeds every section of your business, vendors, clients, jobs, equipment, team, certifications, and more."

Two smaller tiles sit beside the big number:

• Signals Fed, the count of documents that triggered platform intelligence, meaning something in that document changed a benchmark, a finding, or a decision somewhere else in Verinode.
• Needs Review, documents where Verinode could not confidently match the sender to a vendor, carrier, or job on file. Zero here reads as "all matched."

At the bottom of the panel, one line states plainly how the Vault is secured. This is not marketing copy, it is the accurate technical claim about how your data is protected at rest.

The Explore row: six tiles into the same Vault

Below the hero sits a row of tiles labeled Explore. Each one opens the slider directly on its matching tab.

• Uploads. The count is your total document count. The sub-label rotates: how many need review, how many are pending, or "Documents Captured" if neither applies. The small chart under the number splits your documents into signals-fed (green), needs-review (amber), and pending (neutral), so you can read the health of your pipeline at a glance before you open anything.
• Tracking. The count is how many distinct entities, vendors, clients, jobs, Verinode is actively tracking documents against. The sub-label shows the average extraction completeness across those entities, or "Add Data To Unlock" if nothing has flowed in yet.
• Playlist, Reading, Saved. These are your three personal lists, carried over from cards you saved out of the Feed. Playlist and Reading each show how many saved items are still unlistened or unread; Saved is a flat count of everything you bookmarked. They are not invoice-related, but they share the same Vault home because they are also things Verinode is holding for you.
• Notifications. The count defaults to your unread count (or the total if everything is read). The sub-label reads "All Caught Up" when there is nothing pending, or notes how many are unread and, if relevant, how many are scheduled for later.

Recent uploads: your last eight captures

Under Explore sits a row titled Most recent, showing up to eight of your latest documents as individual tiles. Each tile shows the document type (invoice, certificate, estimate, and so on, humanized, never the raw database label), the vendor, carrier, or job it is tied to, the dollar amount if one was extracted, how long ago it was processed, and, when the document triggered platform intelligence, a copper "N signals" badge. A colored completeness badge in the corner shows the extraction score for that specific document: green at 80% or higher, amber from 40 to 79%, red below 40%.

Click any tile and the Uploads tab opens with that exact document expanded, so you land straight on its detail instead of having to find it again in a list.

Empty state. With no documents yet, the row reads: "Recent captures appear here as documents flow in. Forward an email, upload a file, or paste a job export, the latest 8 land in this row."

Uploads: the full document table, invoices included

This is where an invoice actually lives day to day. Open it from the Uploads tile, from any recent-capture tile, or from the "Uploads" tab inside the slider.

Filter bar. A search box (search by vendor, carrier, job, or document type), a bucket dropdown (All buckets, Vendors, Clients, Jobs, each with its own count), and a document-type dropdown listing every type present in your data, humanized (so invoice reads "Invoice", not the raw token).

The table, one row per document, with these columns:

• Source, the vendor, client, or job the document is tied to, linked through to that entity's own page, with the bucket (Vendor / Client / Job) underneath. If Verinode auto-created the entity but could not confidently match it, an amber Assign vendor / Assign carrier / Assign job badge sits next to the name. Click it to open the correction panel and point the document at the right entity.
• Type, the humanized document type: Invoice, Certificate, Estimate, Renewal Notice, Statement, Assignment, Payment Notice, Job Export, Carrier Scorecard, Incident Report, Other, and more, never the underlying snake_case value.
• Amount, the extracted total. A dash means no dollar amount was found on that document, which is normal for a certificate or an incident report and worth a second look on an invoice.
• Processed, the date Verinode finished reading the document.
• Signals, how many downstream signals that document fed. A number here is a link through to Signals; a dash means it has not (yet) triggered anything.
• Status, a colored dot and label: Processed (green), Pending (amber, still being read), or Failed (red). Next to it, when Verinode has an extraction receipt for the document, a percentage badge shows completeness, colored the same green/amber/red bands as everywhere else in the Vault. Hovering it tells you plainly what it means: "We captured most of this document," "Some fields are missing," or "We struggled to read this document."

Row actions, on the right of each row: a preview icon (opens the original file if one was attached), a pencil (edit the extracted fields by hand: document type, total amount, period start and end, document date, reference number), and a trash icon (remove the document, with a confirmation that this also removes any signals sourced from it, and cannot be undone).

Clicking the completeness badge opens a plain-English review panel right under the row: a line stating what fields Verinode actually captured from the document, and, if anything came through incomplete or the score is under 80%, a note saying so plus two buttons, Edit fields (jumps to the manual-correction form) and Re-run extraction (re-parses the source file, useful after you replace it with a cleaner copy).

Bulk actions. Check boxes on multiple rows and a floating bar appears at the bottom of the screen with a count and a Remove selected button.

Empty state. With no documents matching the current filters, the table is replaced by an ambient empty state. With genuinely zero documents on file, it reads a variant of "Your document intelligence builds as files flow in." With documents present but the current filter combination matching none, it is a narrower "nothing matches" prompt to clear the search or filter.

Tracking: what Verinode has learned per vendor, client, and job

The Tracking tab answers a different question than Uploads: not "what documents came in" but "what does Verinode now know, and how confident is it, about each entity those documents describe." One row per vendor, client, or job that has at least one document on file.

Columns: Entity (linked to its own page, with a Needs Review flag if unresolved), Type (Vendor / Client / Job), Docs (how many documents feed this entity), Completeness (a colored percentage pill, or "no receipt" if none of its documents carry an extraction receipt yet), Fields (how many distinct data fields have been captured across all its documents), Signals (how many signals this entity's documents have fed, styled copper when nonzero), and Last update (time since the most recent document).

Click a row to expand it and see the actual list of fields Verinode has captured for that entity, each one humanized and shown as a small green chip, for example "Total Amount", "Period Start", "Reference Number." This is the transparency layer: no black box, you can see exactly what Verinode extracted and from where.

A filter bar narrows by entity type, by completeness band (High ≥ 80%, Medium 40 to 79%, Low under 40%, or "No receipt"), and by search.

Empty state. With nothing tracked yet: "Tracking fills as documents flow in and Verinode extracts fields," with three supporting lines explaining that each invoice, certificate, and statement is parsed by capability, that captured fields stack up per vendor, client, or job, and that clicking any row shows exactly which fields are tracked.

Lists and Notifications: the other two tabs

The Lists tab holds anything you saved out of the Feed into your Playlist, Reading list, or Saved (bookmarked) items. It is not invoice data, but it lives in the Vault because it is, like documents, something Verinode is holding on your behalf for later.

The Notifications tab is your full alert inbox: unread, scheduled, and dismissed notifications, filterable by state. Deep-linking ?tab=lists&list=playlist or ?tab=notifications (as the sidebar's own dropdowns do) opens the slider straight into that specific filter.

How to use this section day to day

1. 1Open Vault from the sidebar. Check the hero panel first: a red or amber completeness pill, or a nonzero Needs Review count, tells you something needs your attention before you go looking for a specific invoice.
2. 2Click the Uploads tile (or any recent-capture tile) to see the full document table.
3. 3Filter the Type dropdown to Invoice if you specifically want invoices, or leave it on All types to see everything flowing in from a vendor or job at once.
4. 4If a row is flagged Assign vendor / carrier / job, click the badge and resolve it, that document is not feeding signals correctly until it is matched.
5. 5Click a completeness badge to see exactly what Verinode read from that document. Use Edit fields to correct a wrong amount or date by hand, or Re-run extraction after replacing the source file.
6. 6Switch to the Tracking tab when you want the entity-level view instead: how many documents and how much confidence Verinode has built up about a specific vendor, client, or job over time.

Best-practice example

Say a subcontractor invoice comes in over email. It appears in Recent uploads within minutes, tagged "Invoice," with an amount and a completeness badge. If the badge is amber, click it: the review panel tells you plainly which fields it captured and which it missed, most often a reference number or a period end date on a less standard invoice layout. Fix it with Edit fields rather than deleting and re-sending it, since the signals that invoice has already fed stay attached. Head to Tracking, find that vendor, and confirm the document count and completeness moved in the right direction, that is the concrete proof the invoice is now feeding the rest of the platform correctly.

Related reading

• Forwarding documents, how email forwarding and uploads get documents into the Vault in the first place
• Connecting your data, the other ingestion channels beyond email and manual upload
• Understanding your margin, where invoice totals and vendor costs surface downstream as net income
• Clients and carriers, the entity pages an invoice's Source column links through to
• The Feed, where saved Playlist, Reading, and Saved items originally come from