AI Customer Support Chatbot & Lead Automator

The plugin indexes your published content and answers visitor questions from it, rather than from a model’s general knowledge. You bring your own API key, so there is no per-message fee to us. When the site has no answer, it offers to fetch a human and stores the lead.

See it on WPBay

How it answers

Pages, posts and WooCommerce products are indexed with their titles, text, price and stock. Retrieved passages are fenced into a block the model is told to treat as untrusted data — because anyone who can publish on your site, including an Author or a Shop Manager, could otherwise plant instructions that change what the bot tells every visitor.

It says it is an AI

Article 50 of the EU AI Act has applied since 2 August 2026: a person must be informed they are interacting with an AI, before or at the very beginning. The widget says so at the top of the panel and in the accessible name of the button that opens it. On by default, including on sites that merely update.

Providers

Google Gemini works without a licence and its free tier covers most small sites. A licence adds OpenAI, Anthropic Claude and Groq. Your key is stored with authenticated AES-256 encryption derived from your site salts, and is sent only to the provider you chose.

Free and licensed

Without a licence the plugin is fully functional on Gemini: indexing, retrieval, conversation memory, lead capture and the whole widget. A licence unlocks the other three providers. One download either way.

What it does not do

Visitor messages leave your site — they go to the AI provider you configure, and the readme discloses exactly what is sent. Visitor names, emails and anything typed into the lead form stay on your site. No prompt-level measure fully solves prompt injection; the real boundary is who may publish on your site.

See it on WPBay

Latest writing

• An Author on your WordPress site can rewrite what your AI chatbot tells every visitor

  My chatbot plugin fed indexed site content straight into the model's system prompt with nothing marking it as data. Anyone who could publish — not just an administrator — could change what the bot said. Here is the exploit, the fix, and the counterfactual that proves the fix is the reason it stopped.

• The EU AI Act's chatbot rule started five days ago. My own plugin was breaking it.

  Article 50 of the EU AI Act has applied since 2 August 2026: a person must be told they are talking to an AI, before or at the very beginning. My WordPress chatbot's default greeting told them nothing. What the rule actually says, and the four decisions it took to ship the fix.