Skip to main content

AI Chatbot

Feature Guide

 

Let your visitors ask questions and get instant, AI-generated answers drawn from your site’s own content — with full conversation context and smart filtering.

Overview

WP AI Search Pro includes a floating AI chatbot widget that appears on every page of your site. Visitors type a question, and the chatbot searches your indexed content to deliver an accurate, sourced response — powered by OpenAI or Google Gemini. It maintains conversation context, supports quick-start suggestions, and automatically respects any active search filters.

How to Enable

Activate the chatbot in one step from your WordPress admin:

Settings
AI Search
General Settings
Search Interface

Check the Enable AI Chatbot box and save. Once enabled:

  • A floating chat bubble appears in the bottom-right corner of every page.
  • Chatbot CSS and JavaScript assets are automatically loaded.
  • REST API endpoints for chat operations become active.
  • Conversation history is tracked for the duration of each session.
Note: When disabled, zero additional assets are loaded — there is no performance impact on your site.

What Your Visitors See

 

Chat Bubble

A floating 60×60 px button fixed to the bottom-right corner. One click opens the full chat window.

 

Chat Window

A clean panel with header, scrollable message area, quick suggestions, and an auto-expanding text input.

 

Quick Suggestions

Pre-populated starter questions visitors can click to begin a conversation instantly. Fully customizable.

 

Conversation Memory

Up to 10 messages of context are maintained so follow-up questions feel natural and accurate.

 

Source Attribution

Every answer includes clickable links back to the original posts and pages the answer was drawn from.

 

Accessible

Full ARIA labels, keyboard navigation, and a live region that announces new messages to screen readers.

Supported AI Providers

Choose the provider and model that best fits your needs. The chatbot will fall back to OpenAI automatically if your primary provider is unavailable.

OpenAI

OpenAI Models

  • GPT-4o
  • GPT-4o Mini
  • GPT-4 Turbo
  • O1
Google Gemini

Gemini Models

  • Gemini 2.0 Flash
  • Gemini 1.5 Pro
  • Gemini 1.5 Flash

Key Settings

Setting Default Range What It Controls
Max Response Tokens 1,000 100 – 4,000 Maximum length of each chatbot reply. Higher values allow longer, more detailed answers.
Temperature 0.7 0 – 2 Controls response creativity. Lower values (e.g. 0.2) give precise, factual answers; higher values allow more varied phrasing.
Conversation History 10 messages Customizable Number of prior messages sent as context. Keeps follow-up questions accurate without inflating API costs.

Smart Filter Integration

When your visitors apply filters on the search form — categories, tags, or post types — the chatbot automatically picks them up. Every subsequent answer is scoped to the filtered content, so responses stay relevant.

Example: A visitor selects the “Technology” category in your search form. From that point on, the chatbot only draws answers from Technology posts — until the filter is changed or cleared.

Filter synchronization happens via a lightweight JavaScript event. Both the search form and chatbot must be on the same page. Post-type filtering requires a Pro license.

License Tiers

Feature Free Pro Enterprise
AI Chatbot Widget      
Search-Based Responses      
Quick Suggestions      
Category & Tag Filters —    
Post-Type Filters —    
Custom Taxonomy Support — —  
Conversation Persistence — —  

Performance & Privacy

 

~40 KB Total

Combined CSS + JS footprint. Zero assets loaded when the chatbot is disabled.

 

No Data Stored

Conversations live in the browser only. Nothing is saved to your database. Session ends on page close.

 

2 API Calls / Message

One embedding call for search, one completion call for the response. ~720 tokens per message on average.

Rate Limiting: The chatbot shares the global rate limiter — 50 requests per IP per minute — to prevent abuse. Optional analytics tracking can log queries for 1–365 days (configurable), associated with IP only.

Troubleshooting

  • Verify the Enable AI Chatbot checkbox is checked under Settings → AI Search → Search Interface.
  • Make sure you saved the settings page after enabling.
  • Check the browser console for JavaScript errors that may prevent the widget from rendering.
  • Hard-refresh the page (Ctrl+Shift+R) to clear cached assets.

  • Confirm an API key is configured for either OpenAI or Google Gemini in Settings → AI Search → General Settings.
  • Ensure your site content has been indexed via Tools → Indexing.
  • Check your API provider dashboard for quota or billing issues.
  • Open the browser console and look for failed REST API calls (/wp-json/wp-ai-search/v1/chatbot).

  • Reduce the number of context chunks sent with each query to lower processing time.
  • Check your API provider’s rate-limit dashboard for throttling.
  • Lower the Max Response Tokens setting if answers are longer than needed.
  • Large sites with many indexed posts may benefit from more targeted indexing.

  • Filter display requires a Pro license — verify your license tier.
  • Both the search form and chatbot must be rendered on the same page.
  • Check the browser console for JavaScript errors that may block event listeners.
  • Ensure you’re running the latest version of WP AI Search Pro.

Part of the WP AI Search Pro documentation series

Questions? Visit developer.suspended-domain.com or open a support ticket.