Overview Overview News News

Feature tour: document types

AI Basics 6 min 2026-08-05

Document types as structure for classification and extraction — the base for stable AI pipelines.

What you will learn

  • Understand types as structure
  • Map fields and rules
  • Connect types with agents
Hide transcript Show transcript

Transcript

Meaning

Imagine an invoice arriving in PaperOffice without a label. It is just a file — hard to find, hard to route, and easy to misplace in everyday work.

With the Invoice document type, PaperOffice knows what it is. Teams can search for it with confidence, agents can extract the right fields, and workflows can route it to the correct process without guessing from the filename alone.

The same idea works for contracts, delivery notes, quotes, and receipts. Each document type is a clear everyday meaning — not just a filename sitting in storage without context for people or automation.

That label is the starting point for everything that follows in Document AI.

Why

Here is why that matters in day-to-day work.

An invoice type helps recognition fill supplier, amount, and due date — so checks and payment preparation can start automatically instead of starting from a blank file.

A contract type can trigger different rules than a delivery note. Approvals, retention, and follow-up steps stay aligned with the real document and not with a generic upload.

And retention can be prepared per category — for example longer keeping for invoices than for drafts — so policy travels with the type from the first moment the document is classified and stored.

Idp agent

In the document type editor, a Default IDP Agent can be bound directly to the type.

After AI-DMS automatically recognizes the type — for example Invoice — that agent is queued in addition to the workspace pipeline, with deduplication so the same agent does not run twice for one document.

Classification and extraction then work as one flow: AI-DMS finds the type, and the bound IDP Agent starts without a manual hand-off from the user or a separate trigger step.

Idp capabilities

An IDP Agent does far more than attach a label to a file.

It can extract targeted data from the page — supplier names, amounts, dates, line items, tax IDs, and other fields configured specifically for that document type.

It can run visual checks: confirm that key regions are present, highlight fields on the page, and support review when something looks incomplete, missing, or uncertain.

It can validate values against business rules, prepare human-in-the-loop review when confidence is low, and hand structured results into the next workflow step without retyping data.

So the agent turns recognition into usable data — ready for routing, checks, approvals, and automation across the organization.

Bound to a document type, those capabilities start automatically after AI-DMS recognition.

System own

Start with system types — ready-made labels for common work like invoices and contracts that teams already recognize across regions.

Create an own type only when the business process is truly different — for example a clinic form, a customs pack, or an internal approval package that needs its own meaning and rules.

Filter by region so the catalog matches the jurisdiction without mixing legal requirements from other regions into the same list.

Create

To create a new document type, open Document Types and choose New document type.

Enter a clear display name, a short description for users in lists and tooltips, and an AI hint that explains how the classifier should recognize this type — typically in English, with typical traits and how to distinguish similar documents.

Set the jurisdiction for where the type applies. Optionally bind a Default IDP Agent so extraction and visual checks start after AI-DMS recognition. Configure retention when documents of this type must carry a WORM or retention rule for compliance.

Save the type. It appears in the catalog and can be used in classification and automation immediately — without rebuilding the whole process from scratch.

Use this path only when the business meaning is truly new — not for every minor naming preference.

Retention

A document type can carry a retention rule into real documents as soon as they are classified.

When an invoice is uploaded or recognized, that rule can be applied — including WORM protection for the required period so files stay tamper-proof until the retention window expires.

Always confirm legal bases with specialists before enforcement. PaperOffice helps prepare retention on the type — it does not replace professional legal or compliance advice for your organization.

Abschnitt

Browse the catalog, try searching for Invoice, and start with system types that already match everyday work.

Bind a Default IDP Agent so extraction, visual checks, and validation follow AI-DMS recognition automatically.

Create an own type only for a real business difference, configure retention carefully, then let classification and automation work with clear meaning.

That is how Document Types keep everyday files understandable in PaperOffice AI — from the first upload to long-term retention.

Where would you like to try PaperOffice?

Workspace and Playground are built for the computer. Grids, batch processing, and the API Playground need a wide screen and a keyboard.

Workspace on the computer. On the phone: capture and approve.

Send to yourself by email Open PaperOffice Mobile
Register free Open app PaperOffice App The full product: web, desktop, and mobile. Capture, organize, search, and work on documents with your team. Free account required Open playground Playground Try selected functions immediately — without registration, using a restricted demo API key. No registration, but a restricted demo API key