To enable accurate and efficient data extraction from COAs, the AI Extraction Settings page must be properly configured. This setup ensures that the system can correctly identify suppliers and items while extracting relevant data based on predefined business rules.
How to Configure AI Extraction Settings
Step 1: Set COA Processing Defaults by completing the following:
- Go to Configuration > AI Settings.
- Set Use to 'COA' from the drop down menu.
- Ensure Automatic Data Processing is toggled OFF. This is recommended for initial implementation.
Step 2: Configure Extraction Attributes by clicking Add Attribute.
When an attribute is added to Extraction Attributes on the AI Extraction Settings page, the 'Extract from COA' toggle is automatically enabled in the Attribute Business Rule Header.
The recommended standard attributes are:
- Manufacturing Date
- Supplier Lot ID
- PO Number
If additional attributes need to be extracted, they can be added here.
Extraction Attributes Features
Settings: You can change the Source and/or the Language by clicking on the cog icon.
Accuracy: Click the bullseye icon underneath the Accuracy column to open the Accuracy modal.
- Attribute Matching: You can enter a term/phrase that may be used in the document to represent an attribute.
- Data Transformation Options:
- AI Prompts: You can also transform data using AI Prompts. For example, convert a date from dd/mm/yyyy → mm/dd/yyyy by typing instructions in the prompt. This lets you control how extracted data appears in TraceStation.
- Norm Scripts: Previously a separate column, Norm Scripts are now selected from the modal. After choosing a script, click Validate to ensure it’s ready to use. A green success banner confirms it’s valid.
Processing Settings
Processing Settings control how documents are processed and how the system handles supplier and item identification and AI template creation.
Use the following settings based on your organization's processing requirements:
- Automatic Data Processing: Determines whether successfully extracted data requires manual verification before records are created or updated.
- Automatic Template Creation: Determines whether a new template is created when an existing template cannot be matched.
- Require Exact Matches: Requires exact supplier and item matches rather than allowing similar names, addresses, or item codes to match.
- Enhanced AI Extraction: Allows AI to analyze the complete uploaded document, which can be helpful for images or PDFs with complex layouts.