Everything you'd want to know before you install.
A detailed look at how Product FAQ AI: AI-written product FAQs (Claude or OpenAI) for PrestaShop 1.7.8, 8 and 9 works, why we built it the way we did, and the thinking behind the features above.
An FAQ that answers real questions, with the facts from the product page
Before buying, a customer wonders what is included, how the product should be cared for, whether it suits their needs or which size to choose. These answers are often already on your product page, but scattered. Product FAQ AI rewrites them as short questions and answers, in the language of each product page, using only the data in your catalogue. When the product page does not contain the information, the AI leaves the question out rather than invent an answer.
Review before publication
Nothing is published without you. Each answer is saved as a draft with the list of facts used. The module flags any figure that does not appear on the product page, or any answer that mentions warranty, delivery, returns, certification or health while the product page says nothing about it. You correct, regenerate a single question if you are not happy with it, reorder, then validate.
Anthropic Claude or OpenAI, with the models on your account
The connection wizard tests the key with a free call to the Models API, then saves it encrypted. The model list is the one from your account, with the indicative list price and the estimated cost for 1,000 FAQs. By default: Claude Opus 5.5 with Anthropic and GPT-6.1 Sol with OpenAI; Claude Sonnet 5.5 and Claude Haiku 5.5 are offered to reduce the cost. Calls are billed by the provider to your own account.
Generate an entire catalogue with no surprises on the invoice
Select the languages and the products, then estimate the cost: the module displays the tokens and a range before anything is sent. The queue progresses in short steps as long as the page is open, with no cron job, and stops at the budget you set or at the monthly cap. Economy batch mode costs about half price, with results within 24 hours. The log shows the tokens actually billed, including cache reads.
An accessible, lightweight and understated accordion
The block uses the native HTML elements details and summary: it opens with the keyboard, it is announced correctly by screen readers and it works without JavaScript. The HTML is generated by the server, so the page does not move when it loads, and the stylesheet is loaded only on product pages that display an FAQ. Five styles are provided: minimal, cards, framed, plus buttons and all open.
Structured data: what you need to know
schema.org FAQPage markup is available as an option and is off by default. Google stopped showing FAQ rich results for all sites on 7 May 2026, after restricting them in 2023 to authoritative government and health websites: the markup remains valid but produces no rich results, and the module promises no SEO gain. If DataFirefly All in One SEO already outputs an FAQPage block, no duplicate is added.
Security and privacy
No customer, order or employee data is sent to the provider. The API key is encrypted in the database and is never sent back to the browser. Product page text is treated as untrusted content, and the answer is constrained by a JSON schema then cleaned into plain text. The back office checks the token, the employee's permissions and the shop context for every action.
Upgrading from version 3
Install the 4.0.0 zip over 3.x: your FAQs, their order and their settings are kept. An active FAQ becomes published, an inactive FAQ becomes a draft, a key stored in plain text is encrypted and old model identifiers are replaced with current models.
There are no reviews yet.