paperoffice

n8n nodes for PaperOffice AI: archive documents, extract invoice data, redact PDFs, search your document archive and queue documents for human review.

Package Information

Released: this month
Downloads: 0 weekly / 0 monthly
Latest Version: 0.1.2
Author: PaperOffice AI

Documentation

n8n-nodes-paperoffice

n8n community nodes for PaperOffice AI: archive documents, extract invoice data, redact PDFs, search your document archive and send documents to a human reviewer, from any n8n workflow.

Installation

In n8n, open Settings → Community Nodes → Install, enter @paperoffice-ai/n8n-nodes-paperoffice and confirm. See the n8n guide for community nodes.

Credentials

  1. In PaperOffice AI, open Account → API (app.paperoffice.ai/en/paperoffice-account/api) and create a User token (po_ut_…). A Group token (po_gt_…) or Secret key (po_sk_…) works as well. Publishable keys (po_pk_…) do not work.
  2. In n8n, create a PaperOffice AI API credential and paste the token. n8n tests the key when you save it.

The key is sent only to https://api.paperoffice.ai. Download links that point to another host are refused.

Operations

Document

Operation What it does
Upload Stores a binary file in a workspace. Optional AI-DMS analysis after upload.
Upload From URL Stores a file from a public HTTPS URL.
Get Returns the details of a document.
Get Text Returns the OCR text of a document or of one page.
Download Returns the stored file as binary data.
Search Full-text, hybrid or semantic search in a workspace.
Add Tag Adds a tag to a document.
Check Duplicates Finds documents that may be duplicates.

Document AI

Operation What it does
Extract Data Reads structured data from a stored document or a binary file. Invoice returns invoice number, dates, supplier, VAT ID, net, VAT and total amounts, currency, IBAN, BIC, payment terms and line items. Purchase order, delivery note, receipt or a custom profile (for example invoice:de:DE) return all fields found.
Redact PDF Blacks out or pseudonymizes personal data. Returns the redacted PDF as binary data, named <file>_redacted.pdf. Pseudonymize needs the Ultra model.
Get Job Returns the status and result of a job that was started without waiting.

Extract Data and Redact PDF wait for the result by default (up to 5 minutes, adjustable). Most results arrive in the first answer. For long jobs the node checks the status at growing intervals. If you switch Wait for Result off, the node returns the job ID and you can use Get Job later.

Human Review

Operation What it does
Queue Document Sends a document to a human review (HITL) agent. Reviewers, review fields and the four-eyes rule are set in the agent in PaperOffice AI.
Get Many Agents Lists your review agents.

The reviewers' decision is sent by PaperOffice AI as a signed webhook. Ready-made workflows that verify the signature and act on the decision are in paperoffice-ai/n8n-workflows.

Workspace

Operation What it does
Get Many Lists your workspaces.

Trigger

PaperOffice AI Trigger starts a workflow when a new document arrives in a workspace (polling). The first poll after activation only remembers the newest document; later polls return the documents added since then, oldest first.

Billing

No specific plan is required. Usage is billed to your PaperOffice AI account as usual.

Limitations

  • Files are uploaded in one request; very large files may hit the upload limit of your plan.
  • Search finds documents uploaded seconds before only after indexing.
  • Extract Data returns fields of the page with the most extracted fields. Multi-document PDFs should be split first (PaperOffice AI split, see the AP inbox workflow in the repository above).
  • Queue Document needs an existing review agent in PaperOffice AI (Workflow AI → HITL agents).

Compatibility

Built and tested with n8n 1.x and 2.x (node API version 1). No runtime dependencies.

Resources

License

MIT

Discussion