| 开发者 | fabioplugins |
|---|---|
| 更新时间 | 2026年10月29日 18:02 |
| PHP版本: | 8.0 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
With entry-level models such as GPT-5.6 Luna, the content sent to the AI model for each answer, before the visitor's question and the chatbot's answer are added, costs about 0.30 to 0.33 US cents (roughly 30 to 33 US cents per 100 answers).
The entry-level models from Gemini, OpenAI, Anthropic and Mistral are enough. They give an answer in less than 10 seconds, even when the answer requires cross-referencing information between 1,000 different pages.
We run our tests on a WooCommerce demo site with 1,000 products and on a how-to demo site with 500 pages. We see no difference in results between our tests on websites of 1,000 pages and on websites of fewer than 100 pages.
Yes. In step 1 of the settings page, add a row to "Custom fields to index": choose the content type, pick or type the field key and name it (the name is what the AI reads above the field's text). Fields created with ACF, Meta Box (default storage), Pods (default storage), Toolset and JetEngine (default storage) can be added, and content types created with Custom Post Type UI appear in the menu. Fields stored in a separate database table cannot be read. Once you choose a field for one of your own content types, that content type is scanned in full. Never map private fields: indexed text can appear in chatbot answers. Changes made to a field are picked up by the next automatic or manual scan. The paid version, Fabio ChatBot Biz, also scans your visitors' comments and re-indexes a page automatically as soon as it or its fields change.
Go to the Analytics page. It covers the last 24 hours. You will find the number of questions, the number of unique questions, the most frequent visitor intent, and a list of conversations (one line per visitor). Click a line to read the exchanges and see which links shown by the chatbot were clicked. You can filter by keyword or by intent, explore the word cloud, and download the conversations as a CSV file. The Biz version covers longer periods, up to the full history.
The AI model gives an intent label (purchase, price, delivery, support, comparison, contact, complaint, information, or off-topic) together with each answer, at no extra cost. Conversations recorded before version 1.2.0 have no label: the Analytics page offers an "Analyze with AI" button that labels them in short batches with your own API key. This uses a small number of API requests that you pay to your AI provider.
No cookies and no personal data such as IP addresses, names or email addresses. A random identifier is kept in the browser session storage only, and is deleted when the tab is closed. See the Privacy section for the full list of stored data.