Web-based attribute enrichment uses AI to find public information about a product and suggest values for Attributes in your PIM.
Unlike other AI Configurations, which use information already stored in your catalog, Web-based attribute enrichment searches the live web. Each suggested value includes the source websites used, so you can review the information before saving it.
Use Web-based attribute enrichment for information that is missing from your PIM but is publicly available online.
When to use it
Web-based attribute enrichment works best as a final enrichment option. Use the information already available in your catalog, Assets, shared product data, and supplier documents first.
Common use cases include:
- Filling technical specifications such as weight, dimensions, materials, or power requirements.
- Finding missing brand, manufacturer, model, release date, or country-of-origin information.
- Drafting an original product description from several public sources.
- Enriching newly onboarded or supplier-provided products that contain limited information.
- Matching public information to options that already exist in your catalog.
Web enrichment takes longer than standard AI generation
A live web search usually takes between 10 and 30 seconds per Attribute, compared with approximately 3 to 5 seconds for standard AI generation. It also costs more per call, so consider this before using it for high-volume enrichment.
Supported Attribute types
Web-based attribute enrichment supports:
- Text and Text Area
- Number, Measurement, Date, Yes/No, and Price
- Simple select and Multi select
- Reference Entity link and Asset collection
| Attribute type | How the value is created |
| Number, Measurement, Date, Yes/No, and Price | The AI returns a precise value backed by public sources. |
| Text and Text Area | The AI writes an original synthesis of the information it finds rather than copying a source word for word. |
| Simple select, Multi select, Reference Entity, and Asset collection | The AI matches its answer to options that already exist in your catalog. |
How it works
An administrator first creates a Web-based attribute enrichment configuration. The configuration defines:
- The Attribute to enrich.
- The product information the AI should use as context.
- The instruction the AI should follow.
- Any websites the AI must not use.
Once the configuration is created, you can run it manually with Ask AI or automate it with a Rule or Workflow.
Enrich one product with Ask AI
Run the enrichment
- Open the product you want to enrich.
- Click Ask AI.
- Select the Attributes you want to enrich.
- Select the Web-based attribute enrichment configuration you want to use.
- Run the enrichment and wait while the AI searches the web.
Review the suggestion
Each proposed value appears in a review card.
- Expand Sources + below the suggestion.
- Open the source websites if you want to verify the information.
- Click the check mark to accept the value, or the cross to reject it.
Accepting the suggestion writes the value to the product. Rejecting it leaves the Attribute unchanged.
Every proposed value includes at least one source website.
Automate web enrichment
You can use the same Web-based attribute enrichment configuration in a Rule or Workflow.
| Method | Best used for | Reviewed before saving |
| Ask AI | Enriching individual products manually. | Yes |
| Rule | Updating a defined selection of products automatically. | No |
| Workflow | Enriching products automatically while requiring human approval. | Yes |
Use a Rule to enrich in bulk
- Create or edit a Rule.
- Add the action Attribute value with AI.
- Select your Web-based attribute enrichment configuration.
- Define the conditions that determine which products the Rule applies to.
Rules write values immediately
Values created by a Rule are written to the selected products without a review step. Check the Rule conditions carefully and test it on a small product selection first.
Use Ask AI or a Workflow when a person must approve the values before they are saved.
Add human review with a Workflow
When Web-based attribute enrichment runs inside a Workflow, the proposed values pass through the Workflow review step before they are written to the product.
The reviewer can see the proposed value, open its source websites, and approve or reject the change.
A Workflow that uses Web-based attribute enrichment must include a review step before it can be enabled.
Workflows are available to Enterprise customers.
See the sources after enrichment
After a value has been saved, you can view its sources in the product history.
The product history also shows whether the value was created through Ask AI or a Rule.
When several Attributes are enriched in the same run, the source websites are grouped together for that run rather than shown separately for each Attribute.
Safety and control
Several safeguards are applied before the AI proposes a value.
Every value has a source
A value is only proposed when it is supported by at least one source website. You can open the sources from Ask AI or the Workflow review screen before accepting the value.
Publisher permissions are checked
Before using a website as a source, the system checks the website's published permission files.
If the publisher explicitly prevents AI use of its content, the source is not used and no value is proposed from it. When a website does not publish a preference, it is treated as allowed.
Block specific websites
Each Web-based attribute enrichment configuration has its own blocked-websites list.
The AI is instructed not to use these websites. Its final sources are also checked against the list before the suggestion is returned.
If a blocked website is found, the AI tries the search again. If it still cannot produce a value without relying on that website, the suggestion is rejected.
Creative content is written originally
For Text and Text Area Attributes, the AI creates an original synthesis of the information it finds. It is instructed to rephrase the content rather than copy a source word for word.
Human review
- Ask AI requires you to approve each value before it is saved.
- Workflows require values to pass through a review step.
- Rules write values immediately and do not include a review step.
You remain in control of your content
Web-based attribute enrichment proposes values from public web sources. Your organisation remains responsible for reviewing and publishing the resulting product content.
When no value is returned
Sometimes the AI cannot propose a value. When this happens, an orange warning appears on the Attribute explaining why.
Common reasons include:
- The only usable source is on the configuration's blocked-websites list.
- The source publisher does not allow AI use of its content.
- The AI could not find a sufficiently reliable, source-backed answer.
For example, you may see a message explaining that the only available source is blocked and asking you to remove it from the blocklist or enter the value manually.
When no value is returned, you can:
- Check that the product contains enough identifying information, such as its name, brand, or model number.
- Review the configuration instruction and product context.
- Review the blocked-websites list.
- Enter the value manually when no reliable public source is available.
Availability
Web-based attribute enrichment is available on plans that include access to AI Configurations.