Operating on AI-powered document processing, Document AI actions require specific configurations to extract data, classify documents, or perform intelligent document operations.
Several fields must be specified to configure a Document AI action:
Additionally, optional fields are:
Below are descriptions of all Document AI-related fields and their respective values.
Document AI service #
Select the Document AI service to use. If no services are offered as an option, it means you don’t have AI services enabled in your license.
AI Action Type #
AI Action Type denotes the operation that will be performed on the document. Choose from the following options:
- Extract: Choose this option to extract structured data from documents using AI. The AI Extraction Model and AI Extraction System Instructions fields become available, along with Set Properties configuration.
- Classify: Choose this option to automatically classify documents based on their content. The AI Classifier, AI Classifier System Instructions, and AI Classify Examples fields become available. Additionally, the Confidence Threshold field appears to control classification accuracy requirements.
AI Model Configuration #
The AI model configuration varies depending on the selected AI Action Type. The system automatically determines which model and instructions to use based on your selection of action type.
AI Extraction Model #
Displayed only when AI Action Type is set to Extract.
Specifies the AI model to be used for data extraction operations. Select from available extraction models configured in your environment.
AI Extraction System Instructions #
Displayed only when AI Action Type is set to Extract.
Provide custom system instructions to guide the AI extraction model. These instructions help the AI understand the context and specific requirements for extracting data from your documents.
AI Classifier #
Displayed only when AI Action Type is set to Classify.
Specifies the AI model to be used for document classification operations. Select from available classifier models configured in your environment.
AI Classifier System Instructions #
Displayed only when AI Action Type is set to Classify.
Provide custom system instructions to guide the AI classifier model. These instructions help the AI understand how to categorize and classify documents according to your business rules.
Pages #
This field specifies which pages of the document will be processed by the AI model.
Format: Enter a comma-separated list of page numbers and/or page ranges.
Examples:
1– Process only the first page1,3,5– Process pages 1, 3, and 51-5– Process pages 1 through 51,3-5,7– Process page 1, pages 3 through 5, and page 7
Default behavior: If left empty, all pages in the document will be processed.
Validation: The field accepts only valid page number formats. Invalid formats will be rejected by the system.
Set Properties #
Optional configuration available for Extract action type.
Defines which properties will be populated with extracted data from the document. This collection allows you to map extracted information to specific metadata properties in M-Files.
- Expand the Set Properties section.
- Click on Add Property Mapping.
- Configure the property mapping with the source data field and target M-Files property.
Each property mapping specifies:
- The data field to extract from the document
- The target property where the extracted value will be stored
- Any transformation or validation rules to apply
Create New Object #
Optional configuration available for Extract action type.
Defines new objects that will be created based on extracted document data. This allows the AI to create related objects (such as line items, contact records, or related documents) from information found in the processed document.
- Expand the Create New Object section.
- Click on Add New Object Configuration.
- Define the object type, class, and property mappings for the new object.
Set Files #
Optional configuration available for Extract action type.
Specifies files that will be created, updated, or attached as a result of the AI processing. This can include extracted pages, generated reports, or processed versions of the original document.
- Expand the Set Files section.
- Click on Add File Configuration.
- Define the file source, destination, and any processing rules.
Input Terms #
Optional configuration available for both Extract and Classify action types.
Defines specific terms, keywords, or patterns that the AI should look for or consider during processing. These terms help guide the AI model’s understanding and improve accuracy for domain-specific terminology.
- Expand the Input Terms section.
- Click on Add Input Term.
- Enter the term and optionally provide context or synonyms.
Input terms are particularly useful for:
- Industry-specific terminology
- Custom field names or labels
- Alternative phrasings of important concepts
- Special codes or identifiers
AI Classify Examples #
Displayed only when AI Action Type is set to Classify.
Provides example documents or document characteristics to train and guide the AI classifier. These examples help the system understand what types of documents belong to each classification category.
- Expand the AI Classify Examples section.
- Click on Add Classification Example.
- Define the example characteristics and expected classification.
Each example includes:
- Sample document characteristics or content patterns
- The expected classification result
- Any relevant metadata or context
The more examples you provide, the more accurate the classification will be.
Confidence Threshold #
Displayed only when AI Action Type is set to Classify.
Specifies the minimum confidence level required for the AI to apply a classification. This threshold helps ensure that only high-confidence classifications are automatically applied.
Range: 0.0 to 1.0 (where 1.0 represents 100% confidence)
Default value: 0.8 (80% confidence)
Behavior: If the AI’s confidence in a classification falls below this threshold, the classification will not be automatically applied. This prevents incorrect or uncertain classifications from being set on documents.
Recommendations:
- Higher threshold (0.9-1.0): Use for critical classifications where accuracy is paramount
- Medium threshold (0.7-0.9): Suitable for most general classification scenarios
- Lower threshold (0.5-0.7): Use when you want broader classification coverage and can tolerate some manual review
NOTE! The actual AI model used and system instructions applied are automatically determined by the system based on your AI Action Type selection. When the action type is “Extract”, the system uses the extraction model and instructions; when it’s “Classify”, the system uses the classifier model and instructions.
