Best Practices for Creating AI Document Type Identification
Product: Paige
Best for: Scan Managers, Administrators
Industry: All
Summary: The article outlines best practices for crafting AI guidance that helps Paige separate documents and recognize their type, which improves how it pulls out key information and applies human-like understanding to the content. . By following these guidelines, users can enhance Paige’s flexibility, improve accuracy, and streamline document processing.
Rather than depending on strict formatting or structure, Paige thrives on context that shows how documents are divided, what type they are, and how they should be interpreted.
Providing Paige with examples can be particularly helpful, even if they aren't applicable every time. That's because Paige doesn't use hardcoded logic, formulas, or strict rules. It looks at your documents and interprets them like a human would.
Here are the best practices for providing AI Document Type Identification guidance.
1. Explain the Document’s Purpose and Industry
Specifying the purpose of the document, how it is used, and what industry used it helps Paige better understand the document. Like a person, Paige uses this context to make informed decisions when identifying documents—not just matching patterns, but interpreting meaning.
✅ Example:
“A purchase order (PO) is a document issued by a buyer to a vendor, authorizing the purchase of specified goods or services.”
“A bank statement is a financial document issued by a bank to summarize an account’s activity over a specific period.”
“A W2 tax form is an official document issued by an employer to report an employee’s annual wages and tax withholdings.”
2. List Common Words and Phrases
Use keywords, common terms and phrases to help Paige recognize documents.
✅ Example:
“An invoice includes a unique invoice number, vendor and customer details, invoice date and payment terms.”
“Purchase orders may reference payment terms, expected delivery dates, and the requesting department or buyer's name.”
“W2 forms may include information about Social Security, Medicare, and other benefits. They are formatted according to IRS standards for tax reporting purposes.”
3. Provide Layout Information
Describe the parts or sections that are unique to this document type, along with formatting and the placement of key details.
✅ Example:
“Invoices often feature business branding, such as logos and contact information at the header of the page, and follow a clear, professional layout.”
“Purchase orders usually contain order numbers and vendor details at the top of the page.”
“Contracts contain sections labeled ‘Terms & Conditions’ and ‘Signatures.’”
“Medical prescriptions often have handwritten sections and a doctor's signature.”
4. Give Examples for Variations
Some documents follow a standard structure but may look different across organizations. When describing the document type, give examples of these variations.
✅ Example:
“Some invoices may have a ship-to address and others may not. Some invoices may also include purchase order numbers, while others list only internal reference codes.”
“Employment contracts may vary in layout. Some use bullet points for responsibilities, while others use paragraph format, but all include job title, compensation, and terms of employment.”
“Legal agreements may vary in layout but always contain terms, conditions, and signatures.”
“Meeting minutes might be formal with timestamps and speakers, or informal summaries without exact times.”
5. Differentiate Between Similar Documents
Some documents share similar content but serve different purposes. Providing distinctions helps Paige classify them correctly.
✅ Example:
“An invoice requests payment, while a receipt confirms that payment has been made.”
“A contract is a legally binding agreement, whereas a purchase order is a request to buy goods.”
By incorporating relevant context and examples, you can help Paige more effectively understand and classify documents. Following these suggestions supports smoother processing and helps Paige deliver accurate, consistent results.
Related Articles
Best Practices for Creating AI Extraction Guidance
Product: Paige Best for: Scan Managers, Administrators, Solutions Engineers Industry: All Summary: The article outlines best practices for crafting AI extraction guidance that guide Paige in accurately extracting and interpreting data from documents. ...Updating AI Extraction Guidance for a Field
Product: Paige Industry: All Best for: Scan Manager and Administrators Summary: AI Guidance for Fields This article offers a step-by-step guide to enter or update AI Extraction Guidance for a field. AI Extraction Guidance provides instructions for ...Document Type Permissions
Document Type Permission Document Type Permissions can be configured and assigned to specific Users and Group from the ‘Permissions’ page of the configured Catalog and Document Type. Go to Administration > Catalogs & Document Types > (Specify Catalog ...Document Type Retention Policy
This applies to Content Central Document Management System. Document Type Retention Policy Go to Administration. Go to Catalogs and Document Types. Select a Catalog and Document Type to configure. Click "Retention Policy". In the "Configure Retention ...Document Type Retention Policy
Retention Policy The Retention Policy screen defines the length of time documents of the document type will be retained in Content Central. With an Enabled policy, documents will expire when the document age becomes older than the 'Length of time ...