Generate alt text

Generate Alt Text creates accessible alt-text strings for product images. The model looks at the image directly (via OpenAI’s vision API) and produces a short, descriptive sentence suitable for screen readers and image search.

How to use

Per image, from a product

1. Open a product → Media section.
2. Open the image edit modal in one of two ways:
   • Context menu on an image — right-click any image thumbnail → Edit with AI, or
   • AI assistant button in the section header — opens the modal with image selection.
3. Pick the alt-text generation preset / option and run it. The model returns a short alt string for each selected image.
4. Edit any value you want, then save.

In bulk across products

1. Go to AI Assistant → Jobs → New job.
2. Pick Generate image alt text as the type.
3. Filter the scope (categories, products missing alt, etc.).
4. Pick the AI connection and start the job.
5. Review per-image suggestions on the jobs detail page and Apply (or Accept all not rejected) to write the alt text onto each product_media row.

Configuration

• Custom instructions — e.g. “Always mention the product colour” or “Keep the alt text under 12 words”.
• Language — for inline use, the alt text is generated in your admin language. For bulk jobs, tick multiple target languages to fill alt across all translations in one run.

Tips & gotchas

Tip

The model only sees the image — not the surrounding product text. If you want consistent phrasing (“Product photo: …”), set that as a custom instruction or bake it into a prompt fork.

Caution

Vision calls are more expensive per image than plain text calls. Bulk runs over a 10 000-image catalogue can add up — check the cost estimate before starting a large job.

Related

• Edit image — modify the image itself
• Cost management
• Batch jobs