AI & Business Requirement
During the AI and business requirements phase, product teams complete structured workshop questionnaires to specify AI-powered validation checks for object naming standards, attribute sourcing, reference data, and taxonomy updates, ensuring the platform aligns precisely with operational needs.
How does the PIM Data Prep’s AI Engine work?
- AI engine scans product data from OE catalogs, supplier feeds, engineering docs, ACES/PIES reference tables, customer search logs, and applications to detect gaps, inconsistencies, and errors in attributes and taxonomy that do not meet industry standards.
- PIM Data prep can auto-apply low-risk fixes (e.g., unit normalization, casing, invalid characters) while routing higher-risk changes to humans for review.
What AI validation checks can be done?
- AI for an attribute can be enabled and configured to be required only when an external source update is needed to verify and enrich data.
- The following AI validations checks can be performed:
- Missing attribute values.
- Invalid characters.
- Find and Replace String.
- Derive attribute values.
- Casing all attribute values to either upper case, lowercase, or title case.
- Maximum allowed characters for a string attribute.
- Unit normalization
- Range checks for numeric attribute values.
- Translate from the master country to support localization.
- Spell Check
- Rephrase
- Keyword Search
- Reclassify product classification
- Data Object Naming Standards
Predefined Validation rules to apply to non-PIES attributes
Select the Attribute to apply validation and cleansing rules to non-PIES attributes.
- Attribute value that cannot be null should be flagged as a validation error for correction by the business user or AI, depending on the business action.
- The attribute value must not include the invalid characters specified in the rule condition, and if they are present, they should be removed or replaced according to the business action.
- The attribute value to find the string as specified in the rule condition and replaced according to the business action.
- Derive attribute values from other attributes as per the Rule condition and definition.
- Change the attribute value to upper case.
- Change the attribute value to lowercase.
- Change the attribute value to title case.
- Validate if the attribute definition exceeds the maximum allowed characters; if it does, flag it as a validation error for correction by the business user or AI, based on the business process.
- Split Number & UOM stored as string to Number and UOM to support modern PIM applications for unit of measurement attributes.
- Convert the Attribute value to a standard UOM.
For example, change all UOM values to IN from any other unit of measure. - Validate for numeric attributes that the range is as per the Rule condition and definition. If it does, flag it as a validation error for correction by the business user or AI, based on the business process.
- Translate the attribute value from the master country to support localization.
Predefined Validation rules to apply for reference data
- Provide source to target LOV values.
Source to Target Classification Mapping
- Manually create or import target product classifications and map to the source product classifications.
- Reclassify products to the correct classification.