Automatic document classification

Why classification is essential

Classification gives each document a clear type based on its primary purpose. That type is shown in the Documents list and in document details, so you can quickly scan, sort, and filter.

What it unlocks for you

• Faster filtering in views: narrow large lists by type without opening every file.
• Cleaner handoffs: teammates see the same categories and avoid duplicating review work.
• Better workflows: type-based filters help you group what matters for a claim, audit, or review.

If you need a refresher on import flow and AI extraction, see the information detected by AI on import page.

The 6 categories (overview)

Contradic uses six categories. The AI assigns exactly one category based on the document’s primary purpose.

• Contracts: agreements that create obligations or authority (contracts, NDAs, licenses, leases).
• Changes: updates to an existing agreement (change orders, amendments, renewals, terminations).
• Scoping: early-stage definition work (RFPs, proposals, quotes, draft statements of work).
• Operations: evidence of execution (delivery notes, site logs, inspection reports, progress reports).
• Finance: payment and coverage records (invoices, purchase orders, receipts, guarantees, insurance).
• Records: governance or context (emails, meeting minutes, compliance documents, legal opinions).

If an email includes a key attachment, the attachment’s content takes priority. When two categories seem possible, the AI chooses the most specific primary purpose.

Correct a classification

You can fix a wrong category in two places. Updates are immediate and affect your filters in views.

Option 1: edit directly in the Documents list

1. Open Documents.
2. Locate the Type column.
3. Pick the correct category from the dropdown.

Option 2: edit in the document details panel

1. Open the document.
2. In Document information, find Type.
3. Select the right category.

If a document is hard to classify, you can clear the value and set it later.

Reliability and limits

Classification is reliable for clear, single-purpose documents, but it can be uncertain when content is short or mixed.

When it can be wrong

• Mixed content: a single file that bundles a proposal and an invoice.
• Thin context: short emails, cover letters, or low-text scans.
• Ambiguous drafts: early-stage documents that look like both scoping and contracts.

How to keep results accurate

• Treat the suggested type as a starting point.
• Review types for high-impact documents (master agreements, change orders, claims evidence).
• When in doubt, update the type manually so your views stay trustworthy.

Learn more about AI extraction

See what the AI detects on import