How to configure the Alt Text tool?

The module generates product image alt text from catalog data. These descriptions help explain the visual and its context and must match each image. They do not guarantee AI citation benefits.

1) Generation mode

Go to the "Alternative text" tab of the module. Select one of the two generation modes:

1.1 Economical mode

Economic mode uses product data: name, category and features. It does not require an AI provider call.

1.2 Premium mode

Premium mode enhances the generated text using an AI provider. It does not send the image to a vision model: check that the description matches the visual.

Configure AI Configuration > OpenAI: key for the official OpenAI endpoint, model and API URL. A custom compatible endpoint can be used; authentication depends on that provider. Save and test the connection.

Don’t forget to save.

2) Real-time processing

Real-time regeneration processes queued items in small batches while the shop is browsed. It spreads the workload; monitor progress and use manual processing when needed.

Remember to save.

3) Alternative text manager

When you add new products or edit existing ones (modification of key fields such as name, short description, reference/SKU, etc.), the generation or update of alternative text is scheduled and the affected products are automatically put in the queue (automatic synchronization).

You can also manually fill the queue yourself via the alternative text manager. To do so, go to the Alternative Text Manager tab in the left menu of your PrestaShop back office. In "Bulk actions", click:

On the green button "Add missing to queue" to add to the queue only images that do not yet have alternative text (recommended after a catalog import):
Or on "Add all to queue" to update the alternative texts for all your images:

Once the queue is filled, you can:

Either process it manually by clicking "Process queue now":
Or, if you have enabled the "Real-time regeneration" option (see part 2), let the processing happen automatically in small batches as your customers browse the site (this distributed approach is recommended as it avoids performance issues)