# Smart Data Modeler - Agentic channels

Prepare your product catalog for AI shopping platforms with the Smart Data Modeler.

Agentic mapping helps you prepare your product catalog for AI shopping platforms. The Smart Data Modeler with agentic channels supports OpenAI and Stripe Agentic Commerce Suite. Each platform defines its own set of required and recommended product attributes. Products that don't meet these requirements might not be displayed or recommended by AI agents.

When you use the Smart Data Modeler with agentic channels, it analyzes a sample of your existing catalog, maps your product data to the attribute structure your target platforms require, and generates a platform-ready [Product Type](/urn?urn=ctp%3Aapi%3Atype%3AProductType) for your Project. Where your catalog is missing mandatory fields, the mapping flags them so you can resolve the gaps before going live. You can then transform your full catalog to the new structure and import it into your Project.

**AI use:** This feature uses AI to analyze data you provide.

**Your data:** We do not use your data to train models.

**Access:** Agentic mapping is available to AI Hub and AgenticLift customers. To enable it, submit a [support request](https://commercetools.atlassian.net/servicedesk/customer/portal/22/group/479/create/1436) with your Project key and region.

## When to use agentic mapping

Use agentic mapping when you want to sell through AI shopping platforms and need a Product data model that matches their feed requirements.

Unlike [Smart Data Modeler for traditional channels](/merchant-center/ai-agents/sdm-traditional-channels.md), which generates a schema optimized for your own commerce channels and leaves you to import the product data, agentic mapping both generates the schema and maps and transforms your existing catalog data to match the structure AI agents expect.

## Before you start

Prepare the following before you launch the Smart Data Modeler:

- Export a sample of your existing catalog. Supported formats are `csv`, `xlsx`, `xls`, and `json`. Use the original `xlsx` or `json` export if available.
- Audit your catalog for platform-required fields, such as `brand`, `condition`, `gtin`, `image_link`, and `description`.
- Confirm your Product Variant grouping logic. Variants of the same product must share an `item_group_id`.

## Generate a Product Type and attribute proposal

To generate a Product Type and attribute proposal optimized for AI shopping platforms:

1. In the Merchant Center, go to **Settings > Product types and attributes > Add product type**.
2. In the **Choose your sales channel type** dialog, select **Agentic channels**, then select **Continue to Smart Data Modeler**.
3. Upload a sample of your existing catalog. Supported formats are `csv`, `xlsx`, `xls`, and `json`, up to a maximum of 4 MB.
4. Select the target platforms: **OpenAI**, **Stripe Agentic Commerce Suite**, or **All of the above**.
5. Select **Generate proposal**.

If your future plans include more than one platform, select **All of the above** to avoid catalog changes later.

The Smart Data Modeler supports OpenAI and Stripe Agentic Commerce Suite only. Google Gemini doesn't use the Smart Data Modeler. If you sell through Google, configure that agentic channel directly in AI Hub.

## Review the proposal

The Smart Data Modeler returns a proposal organized into the following sections:

- **Attributes mapped with platform-ready fields**: catalog characteristics from your file mapped to the corresponding platform attribute, with attribute type, level (Product or Variant), and any required actions.
- **Attributes not needed in commercetools**: fields from your catalog that AI shopping platforms can't display in product feeds. You can add these later if you expand to non-agentic channels.
- **Already defined fields in commercetools Products**: native fields, such as `sku`, `name`, `description`, `images`, `categories`, `slug`, and `key`, that don't need to be re-created as custom Product Type Attributes.

### Resolve mapping actions

Some mapped attributes require a decision before you continue:

- **Partially mapped attributes**: when multiple catalog fields map to a single platform attribute, choose how to resolve them. You can merge the fields into a single attribute and prioritize one field over the other, or map only one of the fields to the platform attribute.
- **Missing values**: when a mandatory platform field has no match in your catalog, add a default value to apply to all products.

When you are satisfied with the mappings, select **Confirm list and continue analysis**.

## Import or export the proposal

After you review the proposal and make any adjustments, choose one of the following options:

- **Transform product catalog with this attribute schema**: creates the Product Type and Attributes in your Project, and generates a `json` file with your products mapped to the commercetools data model. You can download this file and [import it through the HTTP API Playground](/merchant-center/developer-settings.md#create-product-types-from-smart-data-modeler-with-http-api-playground) without transforming the products beforehand. You can follow the import process in the import logs.
- **Add attribute schema to project only**: creates the Product Type and Attributes in your Project without transforming your catalog data.
- **Download attribute schema as JSON**: downloads the proposal as a `json` file so you can analyze it later, without creating anything in your Project yet.

If the proposal requires significant revision, select **Restart assistant** to try again with a revised file or a different platform selection.

If you already have a Product Type in your Project with the same key as the one generated by the Smart Data Modeler, you can choose to replace the duplicated Product Type or keep both. Replacing a Product Type might make existing Products invalid.

## Mandatory attributes

Each AI platform defines its own mandatory attributes. A missing mandatory attribute excludes the product entirely from the feed rather than merely disadvantaging it. The following attributes are common across platforms:

| Attribute | Type | Description |
| :--- | :--- | :--- |
| `brand` | String | The product brand name. |
| `condition` | Enum | Product state (`new`, `used`, `refurbished`). |
| `gtin` | String | Global Trade Item Number (UPC/EAN) that uniquely identifies the product. Products without a GTIN can be excluded from AI agent queries. |
| `enable_search` | Boolean | Enables discoverability in the AI feed. |
| `enable_checkout` | Boolean | Enables purchasing through Agentic Checkout. |

For the complete list of mandatory attributes per platform, see:

- **OpenAI**: [Product data](/agentic-commerce/ai-hub/openai/product-data.md) and [Item information](/agentic-commerce/ai-hub/openai/item-information.md)
- **Stripe Agentic Commerce Suite**: [Product data](/agentic-commerce/ai-hub/stripe/product-data.md)

## Next steps

After you create the Product Type and import your catalog data, configure the agentic channels that sync your products to each AI platform.

Configure agentic channels: [Manage agentic channels](/agentic-commerce/ai-hub/agentic-channels.md).

## Related pages

- [Area overview page with navigation](/merchant-center.md)
- [Previous page: Smart Data Modeler - Traditional channels](/merchant-center/ai-agents/sdm-traditional-channels.md)
- [Next page: Smart Data Modeler - Agentic fields](/merchant-center/ai-agents/sdm-agentic-fields.md)