AI Product Content Generator

The OpenAI API, every service with an OpenAI-compatible Chat Completions API (for example OpenRouter, Groq, Mistral, Together, DeepSeek, Google Gemini through its OpenAI-compatible endpoint, or a local Ollama or LM Studio server) and the Anthropic API. You enter the base URL, the model and your key in the settings.

You pay your AI provider for what you use, directly and at their prices. Product texts are short: with a small model, a description typically costs a fraction of a cent. Enter the token prices of your model in the settings to see the cost of every request and to set a monthly budget.

Yes. After the installation the add-on runs in test mode: it writes sample texts without any provider and without cost, so you can try the product tab, bulk jobs, review, history and undo.

Not without your confirmation. Suggestions are only written when you click Apply. The only automatic mode, "Fill empty fields only", writes into fields that are empty and skips the others. Every applied text can be undone from the History.

The add-on sends only the facts of the product (name, categories, features, current texts) and instructs the model to use only these and never to invent specifications, prices, certifications or guarantees. Models can still make mistakes, which is why texts are reviewed before they are applied. Good product features give the best texts.

The key is stored encrypted in your database, never shown again in full (only masked), never written to logs and sent only to the provider you configured.

In the language you are editing in the admin panel. To write texts for another storefront language, switch the content language and generate again.

Requests are retried with increasing waits. Bulk jobs wait when the provider's rate limit is reached and continue by themselves; items that still fail are tried again later and can be retried by hand. A rejected key or a used-up budget pauses the job with a message.

Up to 2,000 products per job; start several jobs for bigger catalogs. Jobs run in short steps in the browser or by cron, so they never hit a PHP time limit, and the requests-per-minute setting keeps them within your provider's limits.

Not in 1.0.0: the marketplace administrators generate texts for all vendors' products. A vendor mode with plan-based limits is planned.

Without a valid license (no key after the 14-day start period, an invalid or revoked key, a domain the license does not cover, or no successful check for 14 days), new AI generations are switched off on the product tab, in bulk jobs and in cron. Texts that were already generated can still be reviewed and applied, and the history with undo and the usage log keep working. Your products are never affected. Once the license is valid again, everything is back.

Open Add-ons → Manage add-ons → AI Product Content Generator: the License card at the top says what is wrong: no key, a key for another add-on, a revoked license, or a domain that the license does not cover. Fix it on the License tab and click Check now. See License.

Yes. Subdomains of your licensed domain (for example staging.example.com) and local development hosts (localhost, private IP addresses, .local, .test) are covered by the same license.

Nothing changes for 14 days: the add-on keeps working and retries every hour. Only if no check succeeds for 14 days is it switched off until the next successful check.

No. The check runs only in the admin panel or from cron, at most once a day, never while customers browse or check out. The add-on itself has no storefront part.