Introduction - Fabric
Product Activation
Product Activation is the first step in Activate. In this step, you define the set of products you want to work with. You can either:
- Import new products using a CSV file
- Select products that are already available in your catalog
This creates the working set of products that you’ll enrich, review, and publish.
Experiments
When defining your product set, you can choose to work with your full catalog or create an Experiment using a subset of products. Experiments let you test how a smaller group of products performs after enrichment before applying changes across your full catalog. For more information, see Experiments.
Defining Your Catalog
To begin, bring products into Activate by importing them from a CSV file. After your products are available in Activate, you’ll continue to enrichment to refine and optimize your product data.
The Activate flow can also be invoked programmatically. Submit a catalog with Upload artifact, create an optimization workflow with Create workflow, and poll progress with Get workflow status.
Taxonomy Mapping
Once your products are available in Activate, Product Agent automatically classifies them using its taxonomy mapping system. This helps Product Agent understand what types of products you’re working with and apply the appropriate enrichment logic. For example:
- A clothing brand might include shirts, pants, and jackets
- A home goods store might include coffee makers, chairs, and kitchen tools
This classification enables Product Agent to:
- Apply the correct enrichment logic, such as sizing for apparel or dimensions for appliances
- Determine which product attributes are required or recommended
- Tailor downstream steps based on product type
For best results, define your catalog structure in Taxonomy before activating products. This ensures Product Agent uses your categories and attributes as the source of truth during classification and enrichment.
Manual Mapping
In most cases, Product Agent classifies products accurately without manual input. If your data includes custom categories or unclear labels, you may need to:
- Confirm or refine suggested classifications
- Assign more specific product types
- Reorganize products that were misclassified