Similarity Threshold

Location: MxChat → Chatbot Tab → Setting #1

Description: Controls how closely a user’s query must match your knowledge database entries. Set between 20-85, where higher values require closer matches and lower values allow broader matches. When a user submits a query, it’s vectorized and compared to your knowledge database. The system scores all entries and retrieves the top 5 matches to provide context to the AI.

Debugging & Testing: Admin users can view which documents are matched and their exact scores on the frontend using the “MxChat Debugger” tab (visible when logged in as admin). This helps you understand how well your knowledge base entries are matching user queries.

Advanced Usage: If using the multi-bot add-on, you can set individual threshold values for each bot. Actions also have their own individual threshold settings that can be configured in their respective settings pages, giving you granular control over matching sensitivity.

Auto-Display Chatbot

Location: MxChat → Chatbot Tab → Setting #2

Description: When enabled, displays your chatbot as a floating widget on every page of your website automatically. This is the recommended method for consistent chatbot availability across your site.

Alternative Shortcode Options: If you prefer manual placement on specific pages, use these shortcodes:

• [mxchat_chatbot floating="yes"]: Displays the chatbot as a floating widget on the page
• [mxchat_chatbot floating="no"]: Embeds the chatbot directly within the page content

Page-Level Control: You can adjust this global setting on individual pages and posts using the “MxChat Settings” metabox. This allows you to hide the chatbot on specific pages when needed while keeping it active site-wide.

Important Technical Note: The auto-display option is strongly recommended over shortcodes. Shortcodes can have display issues due to z-index inheritance from their parent elements, causing the chatbot to appear behind other page elements. The auto-display option allows the chatbot to escape these CSS constraints and display properly at all times.

Contextual Awareness

Location: MxChat → Chatbot Tab → Setting #3

Description: When enabled, the chatbot gains access to all information on your website pages. The system performs live page scanning with each question asked (not just on page load), reading most text content while intelligently filtering out navigation menus, sidebars, and other non-essential elements.

Advanced Context Control: Use data-mxchat-context attributes to provide additional context that isn’t visible to users:

• <div data-mxchat-context="This is our premium product with 5-year warranty">Product Name</div>
• <span data-mxchat-context="Price includes free shipping">$299</span>

How It Works: The page scan runs in real-time when each question is submitted, ensuring the AI always has current page context. Hidden content is generally not read, but the system focuses on meaningful text content. The data attributes give you precise control over what additional information the AI sees beyond visible page content, enabling more accurate and contextually relevant responses.

Enable Streaming

Location: MxChat → Chatbot Tab → Setting #4

Description: Allows you to enable streaming responses for supported models, providing real-time response generation as the AI processes your query. Streaming creates a more interactive chat experience by displaying the response as it’s being generated.

Important Testing Requirement: You must use the “Test Streaming Compatibility” button to verify your web hosting environment supports streaming before enabling this feature. If streaming is enabled but your hosting doesn’t support it, you will still receive responses, but markdown formatting and hyperlink URLs will not process correctly.

Supported Models: Streaming is available for all OpenAI, Claude, DeepSeek, and Grok models. Other model providers may not support this feature.

Chat Model

Location: MxChat → Chatbot Tab → Setting #5

Description: Allows you to select your preferred AI chat model from our comprehensive list of supported providers. Once you select a model, the corresponding API key field will appear for configuration.

Available Models:

Google Gemini Models:

• Gemini 2.0 Flash: Next-Gen features, speed & multimodal generation
• Gemini 2.0 Flash-Lite: Cost-efficient with low latency
• Gemini 1.5 Pro: Complex reasoning tasks requiring more intelligence
• Gemini 1.5 Flash: Fast and versatile performance

OpenAI GPT Models:

• GPT-5: Flagship for coding, reasoning, and agentic tasks across domains
• GPT-5 Mini: Faster, more cost-efficient for well-defined tasks and precise prompts
• GPT-5 Nano: Fastest and cheapest; ideal for summarization and classification
• GPT-4.1: Flagship model for complex tasks
• GPT-4o: Recommended for most use cases
• GPT-4o Mini: Fast and lightweight
• GPT-4 Turbo: High-performance model
• GPT-4: High intelligence model
• GPT-3.5 Turbo: Affordable and fast

Claude (Anthropic) Models:

• Claude 4 Opus: Most capable Claude model
• Claude 4 Sonnet: High performance
• Claude 3.7 Sonnet: High intelligence
• Claude 3.5 Sonnet: Intelligent and balanced
• Claude 3 Opus: Highly complex tasks
• Claude 3 Sonnet: Balanced performance
• Claude 3 Haiku: Fastest Claude model

Grok (X.AI) Models:

• Grok 4: Latest flagship model – unparalleled performance in natural language, math and reasoning
• Grok-3: Powerful model with 131K context
• Grok-3 Fast: High performance with faster responses
• Grok-3 Mini: Affordable model with good performance
• Grok-3 Mini Fast: Quick and cost-effective
• Grok 2: Latest X.AI model

DeepSeek Models:

• DeepSeek-V3: Advanced AI assistant

Embedding Model

Location: MxChat → Chatbot Tab → Setting #6

Description: Required setting for selecting an embedding model. MxChat uses RAG (Retrieval-Augmented Generation) technology which relies on vector embeddings to retrieve relevant documents from your knowledge base. Embedding models are responsible for converting text content into numerical vectors for similarity comparison.

Available Embedding Models:

• TE3 Small: Fast and cost-effective embeddings (1536 dimensions, 8K context)
• Ada 2: Balanced performance embeddings (1536 dimensions, 8K context)
• TE3 Large: High-performance embeddings (3072 dimensions, 8K context)
• Voyage-3 Large: Advanced semantic search embeddings (2048 dimensions, 32K context)
• Gemini Embedding: Experimental SOTA embeddings (1536 dimensions, 8K context)

Configuration: Once you select an embedding model, the corresponding API key field will appear below. Both chat and embedding models require valid API keys from their respective providers to function properly.

AI Instructions (Behavior)

Location: MxChat → Chatbot Tab → Setting #7

Description: Controls the chatbot’s behavior and personality by providing instructions that are given to the AI on every interaction. You can adjust the chatbot’s tone, how it reacts to specific scenarios, and include minor information that it should always have access to during conversations.

Usage: The default instructions provided serve as a guide to help you understand the structure and type of information you can include. These instructions shape how the AI responds to users and maintains consistency across all interactions.

Top Bar Title

Location: MxChat → Chatbot Tab → Setting #8

Description: Customizes the title displayed at the top of the chatbot UI on the frontend of your website. By default, this reads “MxChat” but can be changed to any text you prefer to match your brand or website.

Multi-Bot Support: If using the multi-bot add-on, each individual bot can have its own unique title, allowing you to differentiate between different chatbot purposes or departments.

AI Agent Text

Location: MxChat → Chatbot Tab → Setting #9

Description: Controls the text displayed at the top right of the chatbot UI, indicating what mode the chatbot is currently in. This is crucial for user clarity since users can switch between AI chatbot and live agent modes.

Important Note: While you can customize this text, it’s essential to be clear to users whether they’re communicating with an AI agent or a live human agent to maintain transparency and set appropriate expectations.

Require Email To Chat

Location: MxChat → Chatbot Tab → Setting #10

Description: When enabled, displays a form that users must complete before accessing the chatbot. This feature helps with lead generation and user identification.

Additional Fields (when enabled):

• Require Email Chat Content: HTML content shown above the form (default: “Welcome to Our Chat! Let’s get started. Enter your email to begin chatting with us.”)
• Require Email Chat Button Text: Text displayed on the submit button (default: “Start Chat”)
• Require Name Field: Toggle to also require users to enter their name along with email
• Name Field Placeholder: Placeholder text for the name field (default: “Enter your name”)

Behavior Notes: Logged-in users won’t see this form. For testing purposes, use a private/incognito browser tab when logged in as admin. Once a user submits their information, they won’t be asked again for 24 hours to prevent repeated requests while navigating your website. The submitted email and name will appear at the top of their chat transcript on the transcripts page.

Introductory Message

Location: MxChat → Chatbot Tab → Setting #11

Description: Sets the first default message shown to all users when they open the chatbot. This message helps set expectations and guides users on how to interact with your chatbot effectively.

Multi-Bot Support: If using the multi-bot add-on, this can be customized for each individual bot, allowing for department-specific or purpose-specific welcome messages.

Input Copy

Location: MxChat → Chatbot Tab → Setting #12

Description: Sets the placeholder text displayed in the chat input field where users type their messages. This text appears when the input field is empty and helps guide users on what they can ask.

Default Value: “How can I assist?” but can be changed to any text that fits your brand voice or provides better guidance to your users.

Chat Teaser Pop-up

Location: MxChat → Chatbot Tab → Setting #13

Description: Sets the message displayed to users above the floating widget before they open the chatbot. Use this to provide a friendly greeting, instructions, or encourage engagement with your chatbot.

User Experience: Once a user dismisses this pop-up, it won’t appear again for 24 hours to prevent repeatedly showing the same message as they navigate your website.

Testing Note: Use private/incognito browser sessions for testing, as dismissing the pop-up once will hide it for the full 24-hour period.

Toggle Privacy Notice

Location: MxChat → Chatbot Tab → Setting #14

Description: When enabled, displays a privacy notice directly below the chat widget input field. This helps with GDPR compliance and transparency about data usage.

Content Options: You can enter custom privacy policy text and include HTML links to your full privacy policy or terms of service. The notice remains visible at all times during chat interactions.

Enable Complianz

Location: MxChat → Chatbot Tab → Setting #15

Description: Integrates with the Complianz plugin to apply consent logic to your chatbot. This ensures GDPR compliance by respecting user consent preferences.

Requirements: The Complianz plugin must be installed and active on your website for this feature to function.

Behavior: When enabled, the chatbot widget will not display at all unless the user has given proper consent through the Complianz plugin interface.

Open Links in a New Tab

Location: MxChat → Chatbot Tab → Setting #16

Description: Controls how links generated by the chatbot are opened. When enabled, all links will open in new browser tabs, keeping users on your site while allowing them to explore referenced content.

Default Behavior: Links open in the same tab unless this setting is enabled.

Enable Chat Persistence

Location: MxChat → Chatbot Tab → Setting #17

Description: Maintains chat history when users navigate between pages or return to your site within a 24-hour period. This creates a seamless conversation experience across browsing sessions.

User Experience: Users can continue their previous conversations without losing context, improving the overall chatbot experience and reducing repetitive questions.

Quick Questions

Location: MxChat → Chatbot Tab → Setting #18

Description: Adds quick action buttons to the chat UI, displayed right above the user input field. These buttons allow users to quickly ask common questions with a single click.

Functionality: You can add as many quick questions as needed. When clicked, the button automatically sends the message to the chatbot for a response. After clicking, the quick question menu automatically collapses, but users can expand it again using the up arrow.

User Interface: This feature helps guide users toward common inquiries and reduces typing, improving engagement and user experience.

Rate Limits Settings

Location: MxChat → Chatbot Tab → Setting #19

Description: Set message limits for each user role on your website and customize the experience when users reach those limits. The system automatically loads all existing user roles and allows you to configure individual limits for each.

Available Placeholders: You can use these placeholders in your custom messages:

• {limit} – The maximum number of messages allowed
• {timeframe} – The time period for the limit
• {count} – Current number of messages used
• {remaining} – Number of messages remaining

Markdown Links Support: Custom messages support clickable links using markdown syntax:

• [Link text](https://example.com) – Creates a clickable link
• [Visit our pricing](https://example.com/pricing) – Link with custom text
• Plain URLs like https://example.com will automatically become clickable

Toolbar Settings

Location: MxChat → Toolbar & Components Tab → Section #1

Description: Controls the chat toolbar functionality and document upload capabilities. The toolbar can be enabled to display below the chatbot input field, providing users with PDF and Word document upload options. You can individually control the visibility of PDF upload and Word upload buttons based on your needs.

Document Upload Features: When files are uploaded, their information is added to the chatbot’s context for answering questions. File names appear at the bottom of the chatbot UI so users know they’re available, and can be removed with the X button. You can customize the intent trigger text (shown when PDF intent is triggered), success text (displayed when vectorization is complete), and error text (shown when processing fails).

Performance Control: The Maximum Document Pages setting (1-69 pages) limits upload size to maintain performance and ensure accurate AI responses. This restriction applies to both PDF and Word documents.

Loops Settings

Location: MxChat → Toolbar & Components Tab → Section #2

Description: Integrates with Loops for automated email marketing campaigns. Enter your Loops API Key and refresh the page to load available mailing lists. Users will be added to your selected list when the Loops action is triggered and an email address is successfully collected.

Smart Email Capture: Configure how the AI approaches email collection with the Triggered Phrase Response (guides the AI on naturally asking for emails when action is triggered) and Email Capture Response (instructions for confirming email capture). The system intelligently captures emails even if users don’t provide one immediately – if they ask questions and later provide an email, it will still be captured.

Campaign Integration: You can set up automated campaigns for specific lists within the Loops platform, enabling automatic email sequences once users are added to your selected mailing list.

Brave Search Settings

Location: MxChat → Toolbar & Components Tab → Section #3

Description: Enables web search and image search functionality through Brave’s API. Enter your Brave Search API Key to activate these features. Configure the number of images to return (1-6) for image searches, which are structured and formatted nicely within the chatbot UI.

Search Configuration: Set Safe Search levels (“Strict” or “Off” – only options supported by Brave), number of news articles to retrieve (1-10), country code for localized results (e.g., “us”), and language code for search results (e.g., “en”). These settings ensure search results are appropriate and relevant to your target audience.

Integration: Both Brave image search and news search actions can be configured in the Actions menu page to trigger based on specific user queries and keywords.

Live Agent Settings

Location: MxChat → Toolbar & Components Tab → Section #4

Description: Comprehensive Slack integration for live agent support. Control agent availability with Live Agent Status (Online/Offline), set notification messages for live transfers, and configure away messages when agents are unavailable. This section also contains all the configuration settings needed to connect your MxChat chatbot with your Slack workspace.

Setting Up the User-to-Agent Slack App:

Follow these steps to create and configure the User-to-Agent Slack app:

1. Download and Edit the Manifest

• Download the website_to_slack file: Download website_to_slack Manifest (if you previously had Slack setup, you need to download this new manifest).
• Open the file in a text editor (e.g., Notepad).
• Replace example.com with your website’s domain (e.g., yourdomain.com).
• Save the changes.

2. Import the Manifest into Slack

• Visit api.slack.com/apps.
• Click “Create New App” (or select your existing MxChat app if updating).
• Select “From an app manifest.”
• Choose your workspace and paste the contents of the edited manifest file as a YAML.
• Click “Next” and then “Create” (or “Update” if modifying an existing app).

3. Gather Required Credentials

After creating/updating the app, collect the following details and enter them in the MxChat Live Agent Settings:

• Signing Secret: Found under “Basic Information” in the app settings. Paste it into the “Slack Secret Key” field in the MxChat settings.
• Bot User OAuth Token: Found under “OAuth & Permissions” after installing the app (starts with xoxb-). Paste it into the “Slack Bot OAuth Token” field in the MxChat settings.
• Webhook URL: Found under “Incoming Webhooks” after activating webhooks. Paste it into the “Slack Webhook URL” field in the MxChat settings.

4. Configure Event Subscriptions NEW

This new feature enables automatic message forwarding between Slack channels and your website chat:

• In your Slack app settings, navigate to “Event Subscriptions” in the left sidebar.
• Toggle “Enable Events” to ON.
• In the “Request URL” field, enter: https://yourdomain.com/wp-json/mxchat/v1/slack-messages (replace with your actual domain).
• Click “Retry” until you see “Verified ✓”.
• Under “Subscribe to bot events”, ensure message.channels is listed.
• Click “Save Changes”.

5. Configure Slack Agent User IDs NEW

Set up which Slack users should be automatically invited to new chat channels:

• Find Your Slack User IDs:
  • In Slack, click on a user’s profile picture or name.
  • Click “View profile”.
  • Click “More” → “Copy member ID”.
  • The ID will look like U1234567890.
  • Repeat for each agent who should receive chat notifications.
• Add to MxChat Settings:
  • In your WordPress admin, go to MxChat → Toolbar & Components → Live Agent Settings.
  • Find the “Slack Agent User IDs” field.
  • Enter one User ID per line (e.g., U1234567890, U0987654321).
  • Save your settings.

6. Configure Live Agent and Switch to Chatbot Actions

To enable seamless switching between live agents and the chatbot, configure the following actions in your MxChat settings:

• Live Agent Action:
  • Go to MxChat → Actions.
  • Click “Add Action” and give it a label of your choice, such as Live Agent.
  • Enter key phrases that should trigger this action (e.g., talk to a person, need help from an agent, live support), separated by commas.
  • In the dropdown menu labeled Action, select Live Agent.
  • Save the action.
• Switch to Chatbot Action:
  • Go to MxChat → Actions.
  • Click “Add Action” and give it a label of your choice, such as Switch to Chatbot.
  • Enter key phrases that should trigger this action (e.g., back to bot, no longer need agent, use chatbot), separated by commas.
  • In the dropdown menu labeled Action, select Switch to Chatbot.
  • Save the action.

These actions allow users to easily transition between live agents and the chatbot during a conversation, ensuring a seamless experience.

MxChat Live Agent Settings Configuration:

After completing the Slack app setup above, configure the following settings in MxChat → Toolbar & Components → Live Agent Settings:

• Live Agent Status: Set to “Online” to allow live agent handovers. Set to “Offline” to disable live agent functionality.
• Notification Message: The message displayed to the user when the live agent has been notified (e.g., “Live agent has been notified. Someone will be with you shortly.”).
• Away Message: The message displayed to the user when live agents are unavailable (e.g., “Sorry, live agents are currently unavailable. Please try again later.”).
• Slack Webhook URL: Enter your Slack Webhook URL (obtained in step 3 above).
• Slack Secret Key: Enter your Slack Signing Secret (obtained in step 3 above).
• Slack Bot OAuth Token: Enter your Slack Bot User OAuth Token that starts with xoxb- (obtained in step 3 above).
• Slack Agent User IDs: Enter the Slack User IDs of agents to invite to chat channels, one per line (obtained in step 5 above).

Security Note: All Slack credentials must be kept secure. Never share your Bot OAuth Token, Secret Key, or Webhook URL publicly.

Knowledge Import Settings

Location: MxChat → Knowledge → Knowledge Import Tab

Description: The Knowledge Import system allows you to add custom content to your AI chatbot’s knowledge base. Content is vectorized using your selected embedding model and stored either in WordPress database or Pinecone (if enabled). The chatbot uses this knowledge to provide accurate, context-aware responses using RAG (Retrieval-Augmented Generation) technology.

API Key Requirements

Before importing knowledge, ensure you have configured the appropriate API key based on your selected embedding model:

• OpenAI models (text-embedding-ada-002, text-embedding-3-small, text-embedding-3-large): Requires OpenAI API key
• Voyage AI models (voyage-2, voyage-large-2, voyage-3-large): Requires Voyage AI API key
• Google Gemini models (gemini-embedding-001): Requires Google Gemini API key

Success Indicator: Green checkmark with message “We detected your [Provider] API key. Remember to add credits to your [Provider] account before using the knowledgebase.”

Error Messages:

• Important: Before importing knowledge, you must add a [Provider] API key with sufficient credits in the Chatbot settings. – Configure your API key in MxChat → Chatbot settings before importing content.
• Missing [Provider] API key. Please configure your API key in the bot settings. – API key not found for the selected embedding model.
• [Provider] API key not configured for bot: [bot_id] – When using multi-bot add-on, ensure each bot has its own API key configured.

WordPress Content Import

Location: MxChat → Knowledge → Knowledge Import Tab → WordPress Content (First Option)

Badge: Recommended

Description: Import specific WordPress posts, pages, and custom post types to your knowledge base. This method provides the most control over which content gets added and is the recommended approach for building your knowledge base. Content imported here is automatically tagged as “In Knowledge Base” for easy identification.

How It Works:

1. Click the “WordPress Content” button to open the content selector modal
2. Use filters to narrow down content:
   • Search: Find content by title or excerpt
   • Content Type: Filter by post type (Posts, Pages, Products, Custom Post Types, Forum Topics, etc.)
   • Status: Filter by publish status (Published, Draft, All Statuses)
   • Processing Status: Show all content, only content already “In Knowledge Base”, or only unprocessed content
3. Select content using checkboxes (or use “Select All” for bulk selection)
4. Click “Process Selected Content” to import to knowledge base

Special Features:

• ACF Field Support: Automatically includes Advanced Custom Fields data in the knowledge base, extracting meaningful text from complex field types
• Custom Post Type Support: Works with WooCommerce products, bbPress forums, wpForo topics, job listings, and other custom post types
• Update Detection: Re-importing content updates existing entries rather than creating duplicates
• Bulk Processing: Process multiple items efficiently with batch operations (one at a time to avoid request size issues)

Content Included:

• Post/page title and main content (with WordPress filters applied)
• Custom fields and meta data
• For job listings: Location, job type, company name
• For products: Product details, pricing, SKU
• For custom post types: Relevant taxonomy terms and metadata

Error Messages:

• No content selected – Select at least one item before clicking Process Selected Content.
• Post not found – The selected post may have been deleted. Refresh the page and try again.
• Storage failed: [error] – Issue storing content. Check API key configuration and database connection.
• Unauthorized access – Requires administrator permissions to process content.
• Invalid nonce – Security check failed. Refresh the page and try again.
• [Provider] API key not configured – Add your embedding API key in Chatbot settings.

Sitemap Import

Location: MxChat → Knowledge → Knowledge Import Tab → Sitemap Import (Second Option)

Description: Import multiple pages from your website using a sitemap XML file. Use a content-specific sub-sitemap rather than your main sitemap index for best results. This method allows bulk importing of website content but requires careful sitemap selection.

How It Works:

1. Click “Sitemap Import” to reveal the URL input field
2. Enter your sitemap URL (e.g., https://example.com/post-sitemap.xml)
3. Click “Import” to begin processing
4. Processing runs in the background – you can monitor progress in the status card
5. The system processes URLs in batches (default: 10 URLs per batch with 5-second pause)

Important Notes:

• Use Sub-Sitemaps: Submit content-specific sitemaps (post-sitemap.xml, page-sitemap.xml) rather than the main sitemap index
• Background Processing: Large sitemaps process in batches to prevent timeouts
• Retry Logic: Failed URLs are automatically retried up to 5 times with exponential backoff
• Content Extraction: The system intelligently extracts main content while filtering out navigation, sidebars, and non-essential elements

Processing Status Card:

When processing is active, a status card displays:

• Progress bar showing completion percentage
• Current progress (processed URLs / total URLs)
• Failed URLs count (if any)
• Last update timestamp
• “Stop Processing” button to cancel the operation
• “Process Batch” button to manually trigger the next batch if processing stalls

Error Messages:

• Please provide a valid URL. – URL field is empty or invalid format.
• Failed to fetch the URL: [error] – Could not access the sitemap URL. Check URL accessibility and network connectivity.
• Empty response received from URL. – The sitemap URL returned no content.
• Invalid sitemap XML. Please provide a valid sitemap. – The XML structure is invalid or malformed.
• No valid URLs found in sitemap – The sitemap contains no processable URLs.
• Embedding API validation failed: [error] – API key test failed before processing began. Check API key configuration.
• Too many batch failures: [error] – Processing stopped after 5 consecutive batch failures.

Manual Batch Processing:

If processing appears stalled (no updates for 5+ minutes), use the “Process Batch” button to manually trigger the next batch of URLs. This is useful if your server’s cron system experiences delays.

Direct URL Import

Location: MxChat → Knowledge → Knowledge Import Tab → Direct URL (Third Option)

Description: Import content from any single webpage by providing its URL. This method is ideal for adding individual pages, blog posts, or external content to your knowledge base.

How It Works:

1. Click “Direct URL” to reveal the URL input field
2. Enter the webpage URL (e.g., https://example.com/about-us)
3. Click “Import” to process the page
4. Content is extracted, vectorized, and added to your knowledge base immediately

Content Extraction:

The system uses intelligent content extraction to:

• Target main content areas using multiple selector strategies
• Support Elementor, Gerow theme, and other page builders
• Filter out navigation menus, sidebars, and footer content
• Preserve meaningful text while removing HTML tags
• Handle various content structures and theme layouts

Status Display:

After submission, a status card shows:

• Success or failure badge
• Submitted URL with link to source
• Submission timestamp (e.g., “2 minutes ago”)
• Content length in characters (on success)
• Embedding dimensions (on success)
• Error details (on failure)

Error Messages:

• No valid content found on the provided URL. – The page content could not be extracted or was empty after processing.
• Failed to generate embedding: [error] – API error occurred during vectorization. Check API key and credits.
• Failed to store content in database: [error] – Database error occurred. Check database connection and table structure.

Direct Content Import

Location: MxChat → Knowledge → Knowledge Import Tab → Direct Content (Fourth Option)

Description: Manually paste or type content to be vectorized and added to your knowledge base. This method is ideal for adding custom information, FAQs, or content that doesn’t exist on a webpage.

How It Works:

1. Click “Direct Content” to reveal the content input area
2. Enter your content in the textarea (up to 10,000 characters)
3. Optionally provide a source URL for reference
4. Click “Import Content” to process

Use Cases:

• Adding frequently asked questions and answers
• Including internal documentation or policies
• Adding product specifications or technical details
• Creating custom responses for specific topics
• Supplementing existing content with additional context

Content Processing:

• Content is sanitized to remove control characters and invalid UTF-8
• Whitespace is normalized while preserving paragraph breaks
• Maximum content length is 65,000 characters (MySQL TEXT field limit)
• Content is vectorized using your selected embedding model

Error Messages:

• API key is not configured. Please add your API key in the settings before submitting content. – Add your embedding API key in Chatbot settings.
• Error storing content: [error] – Database or API error occurred during processing.

PDF Import

Location: MxChat → Knowledge → Knowledge Import Tab → PDF Import (Fifth Option)

Description: Import knowledge from PDF documents by providing a direct URL to the PDF file. The system downloads, parses, and processes the PDF content page-by-page, adding each page to your knowledge base with proper metadata.

How It Works:

1. Click “PDF Import” to reveal the URL input field
2. Enter a direct URL to a PDF file (e.g., https://example.com/document.pdf)
3. Click “Import” to begin processing
4. PDF is downloaded, validated, and page count is determined
5. Processing runs in background with batches of 15 pages at a time (10-second pause between batches)

PDF Processing Features:

• Multiple Parser Methods: Uses Smalot PDF Parser as primary, with fallback to pdfinfo command, PDF repair attempts, and manual structure analysis
• Auto-Repair: Automatically attempts to repair corrupted or malformed PDFs
• Retry Logic: Each page is retried up to 3 times with exponential backoff if processing fails
• Metadata Preservation: Each page includes document type, total pages, current page number, and source URL
• Page Navigation: Metadata includes previous/next page references for context

Processing Status Card:

During processing, a status card displays:

• Progress bar showing completion percentage
• Current progress (processed pages / total pages)
• Failed pages count (if any)
• Last update timestamp
• “Stop Processing” button to cancel
• “Process Batch” button for manual batch triggering if processing stalls
• Completion summary with successful/failed page counts
• Failed pages list with error details and retry counts (expandable)

Error Messages:

• Failed to save PDF file – Could not write PDF to server. Check server permissions and disk space.
• Invalid PDF: Unable to parse or no pages found – PDF could not be parsed by any method. File may be corrupted, encrypted, or not a valid PDF.
• PDF file not found at path: [path] – Temporary PDF file was deleted or moved during processing.
• Empty text on page [number] – The page contains no extractable text. May be an image-only page.
• No valid content after sanitization on page [number] – Page text could not be cleaned or was entirely invalid characters.
• Embedding generation failed: [error] – API error during vectorization. Check API key and credits.
• Database submission failed: [error] – Could not store page content. Check database connection.
• PDF processing appears to be stalled. No updates for over 5 minutes. – Processing has stopped. Use “Process Batch” button or restart the import.

PDF Validation Methods:

The system uses multiple validation approaches:

1. Smalot PDF Parser: Primary parsing method for standard PDFs
2. pdfinfo Command: Fallback system command if available
3. Auto-Repair: Attempts to fix missing PDF headers or EOF markers
4. Manual Analysis: Counts pages by analyzing PDF structure directly

Performance Notes:

• Batch size: 15 pages per batch
• Batch pause: 10 seconds between batches
• Retry attempts: 3 per page with exponential backoff
• Timeout handling: 5-minute stall detection with manual batch trigger option

Knowledge Base Table

Location: MxChat → Knowledge → Knowledge Import Tab → Knowledge Base Section

Description: Displays all content currently in your knowledge base with search, filtering, and management capabilities. The table shows content source, allows inline editing (WordPress DB only), and provides role restriction controls.

Table Columns:

• ID: Unique identifier for WordPress DB entries, or sequential number for Pinecone entries
• Content: The actual text content stored in the knowledge base (read-only for Pinecone, editable for WordPress DB)
• Source: Link to the original URL where content was imported from (shows “N/A” if no source)
• Vector ID: (Pinecone only) Displays the first 12 characters of the vector ID
• Actions: Role restriction dropdown, edit/save buttons (WordPress DB), delete button

Search & Filtering:

• Search Bar: Search through content and source URLs in real-time
• Pagination: 10 entries per page with page navigation
• Record Count: Shows total number of entries with data source badge

Data Source Indicators:

• Pinecone Badge: Blue cloud icon indicates content stored in Pinecone vector database
• WordPress DB Badge: Purple database icon indicates content stored in WordPress database
• Recent Entries Banner: (Pinecone only) Appears when database has 1000+ entries, showing that only recent 1K entries are displayed for performance

Refresh Requirements:

• Pinecone: Wait 60 seconds after adding/deleting content, then refresh page. Banner message: “Refresh page to see new content. New entries or deletions can take up to 60 seconds to reflect! Refreshing too early means content won’t show and you’ll need to refresh again.”
• WordPress DB: Wait 5-10 seconds after adding/deleting content, then refresh page. Banner message: “Refresh page to see new content, but wait 5-10 seconds first! Refreshing too early means content won’t show and you’ll need to refresh again.”

Inline Editing (WordPress DB Only):

1. Click the edit icon (pencil) next to any entry
2. Content and source URL fields become editable
3. Modify the text as needed
4. Click the save icon (checkmark) to save changes
5. Content is re-vectorized with updated text automatically

Delete Operations:

• Single Entry: Click trash icon to delete individual entries (confirmation required)
• Delete All: “Delete All” button removes all knowledge base entries for current bot (confirmation required)
• Pinecone Deletion: Uses AJAX for instant feedback without page reload
• WordPress DB Deletion: Traditional page reload after deletion

Multi-Bot Support:

If using the multi-bot add-on, a bot selector appears at the top allowing you to:

• Switch between different bot knowledge bases
• View “Select Bot Database” dropdown with all configured bots
• See which bot’s knowledge base you’re currently viewing
• Selection is saved automatically and persists across page loads

Error Messages:

• Are you sure you want to delete all knowledge? This action cannot be undone. – Confirmation prompt for Delete All operation.
• Are you sure you want to delete this entry? – Confirmation prompt for single entry deletion.
• Failed to delete entry: [error] – Deletion failed. Check database connection or API credentials.
• Pinecone is not properly configured. – Pinecone settings are invalid or incomplete.
• Invalid data. – Inline edit failed due to invalid or empty content.
• Database update failed. – Could not save edited content to database.
• Embedding generation failed. – Could not generate new vector for edited content.

Role Restrictions

Location: MxChat → Knowledge → Knowledge Import Tab → Knowledge Base Table → Actions Column

Description: Control which users can access specific knowledge base entries based on WordPress user roles. This feature enables you to create public-facing content alongside member-only or administrator-only information within the same knowledge base.

Available Role Options:

• Public (Everyone): Default setting – accessible to all users including non-logged-in visitors
• Logged In Users: Accessible only to users who are logged into WordPress
• Subscribers & Above: Accessible to Subscribers, Contributors, Authors, Editors, and Administrators
• Contributors & Above: Accessible to Contributors, Authors, Editors, and Administrators
• Authors & Above: Accessible to Authors, Editors, and Administrators
• Editors & Above: Accessible to Editors and Administrators
• Administrators Only: Accessible only to site Administrators

How It Works:

• Each knowledge base entry has a role restriction dropdown in the Actions column
• Changes are saved automatically via AJAX (no page reload needed)
• When users ask questions, the system filters knowledge base results by their role
• Content above the user’s permission level is excluded from search results
• The chatbot only receives context from accessible entries

Hierarchical Access:

The system uses hierarchical role checking:

1. Subscriber = Level 1 (lowest)
2. Contributor = Level 2
3. Author = Level 3
4. Editor = Level 4
5. Administrator = Level 5 (highest)

Higher-level roles automatically have access to content restricted to lower levels. For example, an Editor can access Subscriber, Contributor, and Author-restricted content.

Storage (Pinecone):

For Pinecone-based knowledge bases, role restrictions are stored in a separate WordPress table (wp_mxchat_pinecone_roles) that maps vector IDs to role restrictions. This allows role-based access control without modifying the Pinecone vector database.

Storage (WordPress DB):

For WordPress database storage, role restrictions are stored directly in the role_restriction column of the knowledge base table, allowing for faster queries.

Use Cases:

• Create member-only knowledge base content (pricing, internal policies)
• Restrict sensitive information to administrators only
• Provide basic information publicly while offering detailed answers to logged-in users
• Build tiered support systems with different content for different user levels

Error Messages:

• Invalid nonce – Security check failed. Refresh page and try again.
• Unauthorized access – Requires administrator permissions to modify role restrictions.
• Invalid entry ID – The knowledge base entry could not be found.
• Invalid role restriction – The selected role is not a valid option.
• Database update failed: [error] – Could not save the role restriction. Check database connection.

Overview & Stats

Location: MxChat → Transcripts → Stats Dashboard

Description: The dashboard displays three key statistics cards at the top of the page, providing a quick overview of your chatbot activity and user engagement.

Statistics Cards:

• Total Chats: Count of unique chat sessions (distinct session IDs). Each new conversation creates a unique session, even from the same user.
• Total Messages: Count of all messages exchanged across all conversations, including both user messages and bot responses.
• Unique Users: Count of distinct users who have interacted with the chatbot, with detailed breakdown showing:
  • Registered users (identified by email address)
  • Guest users (visitors without email addresses)
  • Agent test sessions (test conversations from agent testing personas)

User Identification Logic:

The system intelligently counts unique users by:

• Prioritizing email addresses for registered users
• Using WordPress user IDs for logged-in users without emails
• Using user identifiers for guests (excluding test personas)
• Falling back to session IDs when no other identifier exists
• Filtering out agent test personas (Tech-Savvy User, Detail-Oriented User, Language Learner, etc.)

Test Persona Detection:

Agent test messages are automatically identified and counted separately. The following test personas are excluded from guest user counts:

• Tech-Savvy User
• Detail-Oriented User
• Language Learner
• Casual Browser
• Policy Enforcer
• Researcher
• Loyalty Member
• Gift Buyer
• Parent or Caregiver

Email Notification Settings

Location: MxChat → Transcripts → Notification Settings Button

Description: Configure email notifications to receive alerts when new chat sessions begin. This feature helps you stay informed about user engagement and respond quickly to customer inquiries.

Opening the Settings Modal:

Click the “Notification Settings” button (envelope icon) in the control bar to open the email notification configuration modal.

Available Settings:

• Enable Notifications: Checkbox to turn email notifications on or off. When enabled, you’ll receive an email notification at the start of each new chat session.
• Notification Email: Email address where notifications should be sent. Defaults to the WordPress admin email address if not specified.

Configuration Instructions:

1. Click “Notification Settings” button to open the modal
2. Check “Send email notification when a new chat session starts” to enable
3. Enter your preferred notification email address (or leave default admin email)
4. Click “Save Notification Settings”
5. The page will reload with settings saved

Modal Controls:

• Save Button: Saves settings and reloads the page
• Cancel Button: Closes modal without saving changes
• X Button: Closes modal (top right corner)
• Outside Click: Clicking outside the modal also closes it

Error Messages:

• Please enter a valid email address for notifications. – The email address format is invalid. Enter a properly formatted email address.

Behavior Notes:

Email notifications are sent at the start of each new chat session, not for every message. If notifications are enabled, you’ll receive one email per conversation initiation, allowing you to monitor new user interactions without overwhelming your inbox.

Search & Filter

Location: MxChat → Transcripts → Search Box

Description: Real-time search functionality to quickly find specific conversations, users, or message content across all chat transcripts.

How to Search:

1. Type your search term in the search box at the top of the page
2. Search executes automatically as you type (no button needed)
3. Results filter in real-time to show matching transcripts
4. Clear the search box to return to all transcripts

Searchable Fields:

The search looks through the following data:

• Session ID: Unique identifier for each conversation
• User Email: Email addresses of registered users
• User Name: Names provided by users (if name field enabled)
• User Identifier: Guest user identifiers or display names
• Message Content: Text content of all messages in conversations

Search Behavior:

• Case-Insensitive: Search is not case-sensitive (matches “John” and “john”)
• Partial Matching: Matches partial text within fields (searching “smith” finds “John Smith”)
• Auto-Reset: Automatically resets to page 1 when search is performed
• Maintains State: Search term persists during pagination through results

Empty Results:

When no transcripts match your search criteria, the system displays: “No chat history found. Try adjusting your search criteria.” Clear the search or modify your terms to see results.

Transcript Display

Location: MxChat → Transcripts → Main Content Area

Description: Chat sessions are displayed in a structured format showing user information, originating page, clicked links, and complete message history for each conversation.

Session Header Information:

• User Display: Shows email address and name (if provided) in format “[email protected] (John Doe)” or just email/identifier if name not available
• Originating Page: Link to the page where the chat session started (includes page title if available)
• Session Time: Timestamp of the last message in format “Month Day, Hour:MinuteAM/PM” (e.g., “Jan 15, 3:45pm”)
• Clicked Links: Shows up to 3 links the user clicked during the conversation, with “+X more” indicator if additional links exist

Message Display:

Each message in the conversation shows:

• Role Badge: Identifies the sender as “Chatbot”, “User”, or “Agent”
• Message Content: The actual message text with HTML formatting preserved (bold, italics, links, lists, etc.)
• Timestamp: Full timestamp in format “Month Day, Year Hour:Minute AM/PM” (e.g., “January 15, 2025 3:45 pm”)

Message Styling:

• Bot Messages: Styled with “bot-message” class for visual distinction
• User Messages: Styled with “user-message” class, displays user identifier or “User”
• Agent Messages: Styled with “agent-message” class for live agent responses
• Unknown Role: Styled with “unknown-message” class for unrecognized message types

HTML Formatting Support:

Messages preserve the following HTML elements:

• Bold and strong tags: <b>, <strong>
• Italic and emphasis tags: <i>, <em>
• Underline: <u>
• Line breaks: <br>
• Paragraphs: <p>
• Lists: <ul>, <ol>, <li>
• Links: <a> with href, title, and target attributes

Link Click Tracking:

If the URL clicks table exists in your database, the system automatically tracks and displays:

• All distinct URLs clicked during each conversation
• Domain names displayed for clicked links (e.g., “example.com”)
• Chronological order of clicks (earliest to latest)
• Visual indication when more than 3 links were clicked

Originating Page Tracking:

When database columns exist, the system displays:

• Full URL of the page where conversation started
• Page title (if available) or URL as fallback
• Clickable link to open the originating page in new tab

Session Selection

Location: MxChat → Transcripts → Session Headers

Description: Bulk selection system allowing you to select multiple chat sessions for batch operations like deletion. Click session headers to toggle selection.

Selection Methods:

• Individual Selection: Click any session header to select/deselect that specific conversation
• Select All Button: Click “Select All” to select all sessions on the current page
• Deselect All Button: When selected, button text changes to “Deselect All” to clear selections

Visual Indicators:

• Selected Sessions: Session container receives “selected” class for visual highlighting
• Button State: Select All button gains “selected” class when active
• Delete Button: Enabled only when one or more sessions are selected

Selection Behavior:

• Persistent Selection: Selected session IDs are stored in a Set, maintaining selection even when navigating between pages
• Page Navigation: Selection state is restored when returning to a page with previously selected sessions
• Automatic Reset: “Select All” state resets when changing pages, but individual selections persist
• Visual Feedback: Sessions remain visually selected when navigating through paginated results

Usage Notes:

• Select All only affects sessions on the current page (50 sessions)
• To select sessions across multiple pages, visit each page and select desired sessions
• Selection count updates in real-time as sessions are selected/deselected
• Delete button shows disabled state when no sessions are selected

Click Handling:

Session headers are clickable for selection except when:

• Clicking the individual delete button (trash icon)
• Clicking links within the header (originating page link, clicked URLs)

Export Transcripts

Location: MxChat → Transcripts → Export All Chats Button

Description: Export all chat transcripts to a CSV file for external analysis, record-keeping, or importing into other systems. The export includes all conversations regardless of current page or search filters.

How to Export:

1. Click the “Export All Chats” button (download icon) in the control bar
2. System retrieves all transcripts from the database
3. CSV file automatically downloads to your computer
4. Button re-enables after 2 seconds

Export File Details:

• Filename Format: chat-transcripts-YYYY-MM-DD.csv (e.g., chat-transcripts-2025-01-15.csv)
• File Encoding: UTF-8 with BOM for proper Excel compatibility
• Content Type: text/csv
• Data Order: Sessions and messages ordered chronologically

CSV Column Structure:

The exported CSV contains the following columns:

1. Session ID: Unique identifier for each conversation
2. Email: User email address (if provided)
3. User Identifier: User display name or guest identifier
4. Role: Message sender role (user, assistant, agent)
5. Message: Complete message text content
6. Timestamp: UTC timestamp of message

Export Scope:

• Exports ALL transcripts from the database (not limited to current page or search results)
• Includes messages from all roles (users, bot, agents)
• Maintains chronological order within each session
• Sessions ordered by session ID and timestamp

Button States:

• Normal: Button enabled and ready for export
• Loading: Button disabled with “loading” class during export
• Re-enabled: Automatically re-enables after 2-second delay

Error Handling:

• No transcripts found. – Database contains no chat history to export.
• You do not have sufficient permissions to access this page. – Requires administrator role to export transcripts.

Security:

Export functionality is protected by:

• Administrator capability check (requires ‘manage_options’)
• Nonce verification for security
• Direct file download without temporary storage

Delete Transcripts

Location: MxChat → Transcripts → Delete Controls

Description: Delete chat transcripts individually or in bulk. The system provides two deletion methods: individual session deletion via trash icons and bulk deletion via the “Delete Selected” button.

Individual Deletion:

1. Hover over any session header to reveal the delete button (trash icon)
2. Click the trash icon
3. Confirm deletion in the browser prompt: “Are you sure you want to delete this chat session? This action cannot be undone.”
4. Session is immediately removed from the database and display

Bulk Deletion:

1. Select one or more sessions by clicking their headers (or use “Select All” button)
2. Click “Delete Selected” button in the control bar
3. Confirm deletion: “Are you sure you want to delete the selected X chat session(s)? This action cannot be undone.”
4. All selected sessions are removed from the database
5. Selection set is cleared and page refreshes

Delete Button States:

• Disabled: When no sessions are selected (default state)
• Enabled: When one or more sessions are selected
• Processing: During AJAX deletion request

Deletion Process:

• Individual deletions pass single session ID to deletion handler
• Bulk deletions pass array of selected session IDs
• Database removes all messages associated with deleted session IDs
• Page automatically refreshes to current page number after deletion
• Deleted session IDs are removed from selection Set

Confirmation Messages:

• Individual: “Are you sure you want to delete this chat session? This action cannot be undone.”
• Bulk: “Are you sure you want to delete the selected X chat session(s)? This action cannot be undone.”
• No Selection: “Please select at least one chat session to delete.” (appears when attempting bulk delete with no selections)

Success/Error Handling:

• Success Alert: “Success: [success message from server]”
• Error Alert: “Error: [error message from server]”
• AJAX Error: “An error occurred while deleting chat sessions. Please try again.”

Post-Deletion Behavior:

• Current page reloads to show updated transcript list
• If all sessions on current page are deleted, user remains on that page number
• Page may show empty if no remaining transcripts exist
• Statistics at top of page update to reflect new counts

Security:

• Nonce verification required for deletion requests
• Administrator capability check (requires ‘manage_options’)
• Session IDs sanitized before database operations

Pagination

Location: MxChat → Transcripts → Bottom of Page

Description: Navigate through chat transcripts with pagination controls showing 50 sessions per page. The system provides page numbers, previous/next buttons, and displays total session count.

Pagination Display:

• Info Text: “Showing X to Y of Z sessions” (e.g., “Showing 1 to 50 of 150 sessions”)
• Page Numbers: Current page and surrounding pages with ellipsis for gaps
• Previous Button: “« Previous” (shown only when not on first page)
• Next Button: “Next »” (shown only when not on last page)

Page Number Display Logic:

• Shows current page plus 2 pages on each side (5 total pages when possible)
• Displays page 1 if not in the immediate range
• Shows last page number if not in the immediate range
• Uses ellipsis “…” to indicate gaps between page ranges
• Current page button receives “active” class for visual highlighting

Navigation Behavior:

• Click any page number to jump directly to that page
• Click “Previous” to go back one page
• Click “Next” to advance one page
• Page automatically scrolls to top of transcripts area on navigation
• Smooth scroll animation (300ms duration) for better UX

Page Size:

• Sessions Per Page: 50 conversations
• Offset Calculation: (page – 1) × 50
• Total Pages: Ceiling of (total_sessions ÷ 50)

Search Integration:

• Pagination persists search terms when navigating between pages
• Search automatically resets to page 1 when new search is performed
• Page counts update dynamically based on filtered results
• Empty search returns to full pagination of all transcripts

Selection Persistence:

• Selected sessions remain selected when navigating between pages
• “Select All” state resets on page change, but individual selections persist
• Selection checkmarks restore when returning to previously visited pages
• Bulk delete operations work across multiple pages

Example Pagination Displays:

• Page 1 of 10: [1] 2 3 4 5 … 10 Next »
• Page 5 of 10: « Previous 1 … 3 4 [5] 6 7 … 10 Next »
• Page 10 of 10: « Previous 1 … 6 7 8 9 [10]
• Page 3 of 5: « Previous 1 2 [3] 4 5 Next »

AJAX Loading:

• Page transitions load via AJAX without full page reload
• “Loading transcripts…” message displays during page fetch
• Error state shows: “Error loading chat transcripts. Please try again.”
• Response includes HTML, current page, total pages, and session count

What Are Actions?

Location: MxChat → Actions

Description: [Content about what actions are, how they work, and why they’re useful will go here]

Actions Manager & Configuration

Location: MxChat → Actions → Main Dashboard

Description: The Actions Manager provides a centralized interface for creating, organizing, and managing all chatbot actions. Actions are displayed in a grid layout with comprehensive search and filtering capabilities. The interface shows 20 actions per page with pagination controls for larger action libraries.

Interface Components:

• Search Bar: Real-time search through action labels and trigger phrases
• Action Type Filter: Dropdown to filter by specific callback functions (Loops Email Capture, Brave Web Search, etc.)
• Filter Button: Apply selected search and filter criteria
• Add New Action Button: Opens the two-step action creation modal
• Actions Grid: Card-based layout displaying all actions with their configuration details

Creating Actions:

Action creation follows a streamlined two-step modal process. Click “Add New Action” to begin, or click “Create Your First Action” if no actions exist yet. The modal guides you through selecting an action type and configuring its behavior.

Step 1: Select Action Type

The action type selector displays all available callback functions organized by category. Browse through categories like Customer Engagement, Search Features, WooCommerce Features, and more. Each action type card shows:

• Action Icon: Visual identifier with dashicon representation
• Action Name: Clear label for the action type
• Description: Brief explanation of what the action does
• Status Badges: “Pro” badge for premium features, “Add-on Required” badge for actions needing specific add-ons
• Add-on Info: Display of which add-on is required if not installed

Category Navigation: Use category buttons at the top to filter action types by purpose. The “All” button shows all available actions. Click any category to view only actions in that group.

Search Functionality: Type in the search box to filter action types by name or description in real-time. The search works across all categories and instantly updates the displayed cards.

Action Availability: Actions with Pro badges require an active Pro license. Actions with “Add-on Required” badges need the corresponding add-on installed. Clicking unavailable actions displays the upgrade requirement.

Step 2: Configure Action

After selecting an action type, the configuration screen displays with the selected action’s details at the top. A “Back to Action Types” button allows returning to step 1 if you need to change the action type. Configure the following settings:

Action Label:

Purpose: A reference name for your internal use only – this label is not shown to users and serves purely for organization in your Actions Manager. Use descriptive labels to easily identify actions later (e.g., “Newsletter Signup”, “Product Search”, “Live Support Request”).

Trigger Phrases:

Format: Enter comma-separated phrases that should activate this action. The system processes each phrase, generates embeddings, and creates an averaged vector for matching.

How It Works: When users send messages, the system vectorizes their input and compares it against all action trigger phrases. If similarity exceeds the threshold, the action triggers. Multiple phrases help capture different ways users might express the same intent.

Best Practices:

• Include 4-8 varied phrases covering different user expressions
• Use natural language users would actually type
• Cover synonyms and different wordings (e.g., “sign me up, subscribe me, I want to join, add me to newsletter”)
• Avoid overly similar phrases – diversity improves matching accuracy
• Consider user intent variations (formal vs. casual, direct vs. indirect)

Processing: Each phrase generates an embedding vector. Failed phrases are reported in error messages with specific phrase names, allowing you to identify and correct problematic entries. Successful vectors are averaged into a combined vector stored in the database.

Similarity Threshold:

Control: Slider ranging from 10% to 95% determines how closely user input must match trigger phrases to activate the action. The value displays dynamically as you adjust the slider.

Threshold Guidelines:

• Low (10-30%): Action triggers very easily – use for broad, general actions or fallback behaviors
• Medium (40-60%): Balanced sensitivity – catches similar intents while avoiding false positives
• High (70-95%): Requires near-exact matches – use for specific, precise actions where accuracy is critical

Action-Specific Threshold: Unlike the global similarity threshold in chatbot settings, this threshold applies only to this specific action. Each action can have its own sensitivity level, allowing granular control over when different actions trigger.

Bot Assignment:

Purpose: Control which chatbots can use this action. By default, actions are assigned to the Default Bot. If using the Multi-Bot add-on, you can assign actions to specific bots or multiple bots simultaneously.

Selection: Checkbox interface displays all available bots:

• Default Bot: Always shown and checked by default for backward compatibility
• Additional Bots: All bots created through Multi-Bot add-on appear as checkbox options
• Multiple Selection: Check multiple boxes to enable the action across several bots

Behavior: The system ensures “default” is always included in the enabled_bots array for backward compatibility with existing functionality. Custom bots are added to the JSON array when selected.

Managing Actions:

The Actions Manager displays each action as a card containing all essential information and controls. Actions are shown in a responsive grid layout with 20 actions per page. Each action card displays:

• Card Header: Action label with enable/disable toggle switch
• Action Type: Callback function name (e.g., “Loops Email Capture”, “Brave Web Search”)
• Trigger Phrases: Preview of trigger phrases (truncated to 100 characters with intelligent comma handling)
• Similarity Threshold: Percentage value showing current threshold setting
• Assigned Bots: Badge display showing which bots have this action enabled
• Card Footer: Edit and Delete buttons (or special workflow buttons for Form and Smart Recommender actions)

Bot Badges: Each assigned bot displays as a styled badge. The Default Bot shows as “Default Bot”, while Multi-Bot add-on bots display their configured names. If bot names cannot be retrieved, the bot ID is shown as fallback.

Phrase Truncation: Long trigger phrase lists are intelligently trimmed to 100 characters. The system cuts at the last comma before the limit and adds “…” to indicate more phrases exist, ensuring clean display without cutting words mid-phrase.

Enabling/Disabling Actions:

Toggle Switch: Each action card header contains a toggle switch for quick enable/disable control. The switch updates via AJAX without page reload, providing instant feedback.

Status States:

• Enabled (Checked): Action is active and will trigger when matching user input is detected
• Disabled (Unchecked): Action is inactive and will not trigger regardless of user input

Behavior: Changes save immediately to the database. If the enabled column doesn’t exist in the database (older installations), actions default to enabled state. The toggle switch uses the mxchat-action-toggle class with data-action-id attribute for JavaScript handling.

Editing Actions:

Edit Button: Click the Edit button on any action card to open the configuration modal pre-populated with the action’s current settings. The modal displays Step 2 (Configure Action) directly, showing the selected action type at the top.

Pre-populated Fields:

• Action label populated in the text input
• Trigger phrases loaded into textarea
• Similarity threshold slider set to current percentage value
• Callback function stored in hidden field
• Enabled bots checkboxes marked according to current assignment

Deleting Actions:

Delete Button: Click the Delete button (trash icon) in any action card footer to remove the action permanently. A confirmation dialog prevents accidental deletion.

Confirmation: Browser prompt displays “Are you sure you want to delete this action?” – clicking OK proceeds with deletion, Cancel aborts the operation.

Permanent Removal: Deleted actions cannot be recovered. The action’s embedding vector, trigger phrases, and all configuration data are permanently removed from the database.

Error Messages:

• Are you sure you want to delete this action? – Confirmation prompt to prevent accidental deletion. Click OK to proceed or Cancel to abort.
• Unauthorized user – User lacks administrator permissions (requires ‘manage_options’ capability). Only administrators can manage actions.
• Invalid input. Please ensure all fields are filled out. – One or more required fields are empty. Check that Action Label, Trigger Phrases, and Callback Function are all provided.
• Please enter at least one valid phrase. – The phrases field contains no valid entries after processing. Enter at least one comma-separated trigger phrase.
• Error generating embeddings for phrases: [phrases] – Embedding API failed for specific phrases listed. Check your embedding API key configuration, ensure you have sufficient credits, and verify the phrases don’t contain problematic characters.
• No valid embeddings generated. – All phrase embedding attempts failed. Verify your embedding API key in Chatbot settings, check API credits, and ensure your embedding model is configured correctly.
• Invalid callback function selected. – The selected callback function doesn’t exist in available callbacks. Refresh the page and select a valid action type.
• This callback function is available in the Pro version only. – Attempting to create a Pro action without an active Pro license. Upgrade to Pro or select a free action type.
• Database error: [error] – Database operation failed with specific error details. Check database connectivity, table structure, and review the specific error message for troubleshooting.

Loops Email Capture

Location: MxChat → Actions → Customer Engagement Category

Description: The Loops Email Capture action seamlessly integrates with your Loops account to automatically add user email addresses to your specified mailing list. This action triggers when a user expresses interest in subscribing or provides their email within the chat, automating your lead generation and email marketing efforts.

Requirements:

• Loops Account: You must have an active account with Loops.so.
• Loops API Key: A valid API key from your Loops account is required. You can find this in your Loops account settings.
• Loops Mailing List: At least one mailing list needs to be created within your Loops account to add subscribers to.
• Configuration in MxChat: Your Loops API Key and desired mailing list must be selected in the MxChat Toolbar & Components → Loops Settings.

Configuration:

1. Navigate to MxChat → Actions and either create a new action or edit an existing one.
2. In the action configuration, select Loops Email Capture as the action type.
3. Enter a descriptive Action Label (e.g., “Newsletter Signup”).
4. Define Trigger Phrases that will initiate the email capture flow (see example phrases below).
5. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. A lower threshold will trigger the action more easily.
6. Assign the action to the desired Bots.
7. In MxChat → Toolbar & Components → Loops Settings, configure the following:
   • Loops API Key: Enter your Loops API key.
   • Mailing List: Select the Loops mailing list to which you want to add subscribers.
   • Triggered Phrase Response: This is the message the chatbot will use to ask the user for their email address. The default message is “Please provide your email address.” You can customize this to match your brand voice.
   • Email Capture Response: This is the success message the chatbot will display after the user provides their email address. The default message is “Thank you for providing your email! You’ve been added to our list.” You can customize this to match your brand voice.

Example Trigger Phrases:

• Sign me up for the newsletter
• Subscribe to your email list
• I want to join your mailing list
• Add me to your newsletter
• Keep me updated with your latest news
• I’d like to receive your emails

Brave Web Search

Location: MxChat → Actions → Search Features Category

Description: The Brave Web Search action allows your chatbot to leverage the Brave Search API to provide real-time web search results directly within the chat interface. When a user asks a question that requires information beyond your knowledge base, this action can be triggered to fetch relevant search results and present them in a concise and user-friendly format. This enhances the chatbot’s ability to answer a wider range of questions and provide up-to-date information.

Requirements:

• Brave Search API Key: A valid API key from Brave Search is required. You can obtain an API key by creating an account on the Brave Search API website.
• MxChat Configuration: The Brave Search API Key must be configured in the MxChat settings under Toolbar & Components → Brave Search Settings.
• Results Count: Configure the number of search results to display in the same settings page.

Configuration:

1. Navigate to MxChat → Actions and either create a new action or edit an existing one.
2. In the action configuration, select Brave Web Search as the action type.
3. Enter a descriptive Action Label (e.g., “Web Search”).
4. Define Trigger Phrases that will initiate the web search. Examples include “search the web for”, “find information about”, “what is”, etc. See example phrases below.
5. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on how specific you want the trigger to be.
6. Assign the action to the desired Bots.
7. Configure the following settings in MxChat → Toolbar & Components → Brave Search Settings:
   • Brave Search API Key: Enter your Brave Search API key.
   • Number of Results: Specify the number of search results to display (default is 5).
   • Safe Search: Choose between “Strict” or “Off” for safe search filtering.
   • Number of News Articles: Set the number of news articles to retrieve.
   • Country Code: Enter the country code for localized search results (e.g., “us” for United States).
   • Language Code: Specify the language code for search results (e.g., “en” for English).

Example Trigger Phrases:

• Search the web for [topic]
• Find information about [topic]
• What is [topic]?
• Can you search for [topic]?
• Look up [topic] on the internet
• I need to know about [topic]

Brave Image Search

Location: MxChat → Actions → Search Features Category

Description: The Brave Image Search action empowers your chatbot to display relevant images directly within the chat interface using the Brave Search API. This allows users to quickly find visual content related to their queries, enhancing engagement and providing a richer, more interactive experience. The images are presented in a visually appealing gallery format.

Requirements:

• Brave Search API Key: A valid API key from Brave Search is required. You can obtain an API key by creating an account on the Brave Search API website.
• MxChat Configuration: The Brave Search API Key must be configured in the MxChat settings under Toolbar & Components → Brave Search Settings.
• Image Count: The number of images to display (between 1 and 6) must be configured in the same settings page.

Configuration:

1. Navigate to MxChat → Actions and either create a new action or edit an existing one.
2. In the action configuration, select Brave Image Search as the action type.
3. Enter a descriptive Action Label (e.g., “Image Search”).
4. Define Trigger Phrases that will initiate the image search. Examples include “show me pictures of”, “find images of”, “I want to see pictures of”, etc. See example phrases below.
5. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
6. Assign the action to the desired Bots.
7. Configure the following settings in MxChat → Toolbar & Components → Brave Search Settings:
   • Brave Search API Key: Enter your Brave Search API key.
   • Number of Images: Specify the number of images to display (1-6, default is 4).
   • Safe Search: Choose between “Strict” or “Off” for safe search filtering.

Example Trigger Phrases:

• Show me pictures of [topic]
• Find images of [topic]
• I want to see pictures of [topic]
• Display images of [topic]
• Pictures of [topic]
• Images of [topic]

Generate Image

Location: MxChat → Actions → Other Features Category

Description: The Generate Image action allows your chatbot to create unique images on demand using the OpenAI DALL-E models. When a user asks for an image of something, this action will trigger, sending a prompt to DALL-E and displaying the resulting image directly within the chat. This adds a fun and creative element to your chatbot interactions.

Requirements:

• OpenAI API Key: A valid OpenAI API key is required. This is the same API key used for the chat model.
• DALL-E Access: Ensure your OpenAI account has access to the DALL-E models.
• MxChat Configuration: The OpenAI API Key must be configured in the MxChat settings under MxChat → Chatbot Tab.

Configuration:

1. Navigate to MxChat → Actions and either create a new action or edit an existing one.
2. In the action configuration, select Generate Image as the action type.
3. Enter a descriptive Action Label (e.g., “Generate Image”).
4. Define Trigger Phrases that will initiate the image generation. Examples include “generate an image of”, “create a picture of”, “draw me a”, etc. See example phrases below.
5. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
6. Assign the action to the desired Bots.

Example Trigger Phrases:

• Generate an image of [subject]
• Create a picture of [subject]
• Draw me a [subject]
• Make an image of [subject]
• Show me a picture of [subject]
• Can you generate an image of [subject]?

Chat with PDF

Location: MxChat → Actions → Other Features Category

Description: The Chat with PDF action enables users to directly interact with and ask questions about PDF documents within the chat interface. Users provide a URL to a PDF, and the chatbot processes the document, allowing them to ask questions and receive answers based on the PDF’s content. This is ideal for providing quick access to information contained in manuals, reports, or other document formats.

Requirements:

• PDF Accessibility: The PDF document must be accessible via a direct URL. The URL should point directly to the PDF file (e.g., `https://example.com/document.pdf`), not a webpage containing the PDF.
• PDF Content: The PDF should contain selectable text. Scanned documents or image-based PDFs may not be compatible unless OCR (Optical Character Recognition) has been applied.
• Maximum Pages: There is a limit to the number of pages that can be processed, configurable in the MxChat settings. The default is 69 pages.
• Configuration in MxChat: The PDF-related settings (intent trigger, success, and error texts) can be customized in the MxChat settings under Toolbar & Components → Toolbar Settings.

Configuration:

1. Navigate to MxChat → Actions and either create a new action or edit an existing one.
2. In the action configuration, select Chat with PDF as the action type.
3. Enter a descriptive Action Label (e.g., “PDF Chat”).
4. Define Trigger Phrases that will initiate the PDF chat. Examples include “read this pdf”, “analyze this document”, “can you chat with this pdf”, etc. See example phrases below.
5. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
6. Assign the action to the desired Bots.
7. Customize the following settings in MxChat → Toolbar & Components → Toolbar Settings:
   • PDF Intent Trigger Text: The message the chatbot will use to ask the user for the PDF URL. Default: “Please provide the URL to the PDF you’d like to discuss.”
   • PDF Intent Success Text: The message the chatbot will display after successfully processing the PDF. Default: “I’ve processed the PDF. What questions do you have about it?”
   • PDF Intent Error Text: The message the chatbot will display if it fails to process the PDF. Default: “Sorry, I couldn’t process the PDF. Please ensure it’s a valid file.”
   • Maximum Document Pages: Sets the maximum number of pages allowed in the PDF.

Example Trigger Phrases:

• Read this PDF
• Analyze this document
• Can you chat with this PDF?
• I want to discuss a PDF
• Process this PDF file
• I have a PDF to analyze

Slack Live Agent

Location: MxChat → Actions → Customer Engagement Category

Description: The Slack Live Agent action allows you to seamlessly hand over conversations from the AI chatbot to a live human agent via Slack. When a user’s request requires human intervention or more personalized support, this action creates a dedicated Slack channel, invites designated agents, and forwards the conversation context, enabling agents to provide real-time assistance directly within Slack. Agent responses are then relayed back to the user in the chat interface.

Requirements:

• Slack Workspace: You need an active Slack workspace.
• Slack App: A Slack App with the necessary permissions is required (setup instructions below).
• MxChat Configuration: Slack credentials must be configured in the MxChat settings under Toolbar & Components → Live Agent Settings.

Setting Up the User-to-Agent Slack App:

Follow these steps to create and configure the User-to-Agent Slack app:

1. Download and Edit the Manifest

• Download the website_to_slack file: Download website_to_slack Manifest (if you previously had Slack setup, you need to download this new manifest).
• Open the file in a text editor (e.g., Notepad).
• Replace example.com with your website’s domain (e.g., yourdomain.com).
• Save the changes.

2. Import the Manifest into Slack

• Visit api.slack.com/apps.
• Click “Create New App” (or select your existing MxChat app if updating).
• Select “From an app manifest.”
• Choose your workspace and paste the contents of the edited manifest file as a YAML.
• Click “Next” and then “Create” (or “Update” if modifying an existing app).

3. Gather Required Credentials

After creating/updating the app, collect the following details:

• Signing Secret: Found under “Basic Information” in the app settings. Paste it into the “Slack Secret Key” field in the MxChat settings.
• Bot User OAuth Token: Found under “OAuth & Permissions” after installing the app. Paste it into the “Slack Bot OAuth Token” field in the MxChat settings.
• Webhook URL: Found under “Incoming Webhooks” after activating webhooks. Paste it into the “Slack Webhook URL” field in the MxChat settings.

4. Configure Event Subscriptions NEW

This new feature enables automatic message forwarding between Slack channels and your website chat:

• In your Slack app settings, navigate to “Event Subscriptions” in the left sidebar.
• Toggle “Enable Events” to ON.
• In the “Request URL” field, enter: https://yourdomain.com/wp-json/mxchat/v1/slack-messages (replace with your actual domain).
• Click “Retry” until you see “Verified ✓”.
• Under “Subscribe to bot events”, ensure message.channels is listed.
• Click “Save Changes”.

5. Configure Slack Agent User IDs NEW

Set up which Slack users should be automatically invited to new chat channels:

• Find Your Slack User IDs:
  • In Slack, click on a user’s profile picture or name.
  • Click “View profile”.
  • Click “More” → “Copy member ID”.
  • The ID will look like U1234567890.
  • Repeat for each agent who should receive chat notifications.
• Add to MxChat Settings:
  • In your WordPress admin, go to MxChat settings.
  • Find the “Slack Agent User IDs” field.
  • Enter one User ID per line (e.g., U1234567890, U0987654321).
  • Save your settings.

6. Configure Live Agent and Switch to Chatbot Actions

To enable seamless switching between live agents and the chatbot, configure the following actions in your MxChat settings:

• Live Agent Action:
  • Go to the “Actions” section of your MxChat settings.
  • Click “Add Action” and give it a label of your choice, such as Live Agent.
  • Enter key phrases that should trigger this action (e.g., talk to a person, need help from an agent, live support), separated by commas.
  • In the dropdown menu labeled Action, select Live Agent.
  • Save the action.
• Switch to Chatbot Action:
  • Go to the “Actions” section of your MxChat settings.
  • Click “Add Action” and give it a label of your choice, such as Switch to Chatbot.
  • Enter key phrases that should trigger this action (e.g., back to bot, no longer need agent, use chatbot), separated by commas.
  • In the dropdown menu labeled Action, select Switch to Chatbot.
  • Save the action.

These actions allow users to easily transition between live agents and the chatbot during a conversation, ensuring a seamless experience.

MxChat Live Agent Settings:

Configure the following settings in MxChat → Toolbar & Components → Live Agent Settings:

• Live Agent Status: Set to “Online” to allow live agent handovers. Set to “Offline” to disable.
• Notification Message: The message displayed to the user when the live agent has been notified (e.g., “Live agent has been notified.”).
• Away Message: The message displayed to the user when live agents are unavailable (e.g., “Sorry, live agents are currently unavailable.”).
• Slack Webhook URL: Enter your Slack Webhook URL.
• Slack Secret Key: Enter your Slack Secret Key (Signing Secret).
• Slack Bot OAuth Token: Enter your Slack Bot OAuth Token (starts with xoxb-).
• Slack Agent User IDs: Enter the Slack User IDs of agents to invite to the channel (one per line).

Example Trigger Phrases:

• Talk to a human
• Connect me with an agent
• I need help from a real person
• Speak to a representative
• Can I talk to someone?
• I need assistance from a human
• Talk to a person
• Need help from an agent
• Live support

Back to Chatbot

Location: MxChat → Actions → Customer Engagement Category

Description: The Back to Chatbot action allows live agents to seamlessly transfer the conversation back to the AI chatbot within Slack. This provides a smooth transition when the agent has addressed the user’s immediate needs or when the AI can handle the remaining tasks more efficiently. This action is crucial for creating a balanced and efficient customer support experience, leveraging the strengths of both human agents and AI.

Requirements:

• Slack Live Agent Action: The Slack Live Agent action must be configured and active.
• Slack App: The same Slack App configured for the Slack Live Agent action is required.
• Slack Configuration: The Slack App needs to have a slash command configured (e.g., `/chatbot`) that triggers this action.

Configuration:

1. Navigate to MxChat → Actions and either create a new action or edit an existing one.
2. In the action configuration, select Back to Chatbot as the action type.
3. Enter a descriptive Action Label (e.g., “Switch to Chatbot”).
4. Define Trigger Phrases. Since this action is triggered by a Slack command, these phrases are primarily for internal documentation. Examples: “switch to chatbot”, “hand back to AI”, “return to AI assistant”.
5. Set the Similarity Threshold. This setting is less critical as the action is triggered by a Slack command, but a reasonable value (e.g., 50%) is recommended for consistency.
6. Assign the action to the desired Bots.
7. Create a Slack Slash Command in your Slack App settings that will trigger this action. For example, you could create a `/chatbot` command.

How it Works:

1. The live agent uses the configured Slack slash command (e.g., `/chatbot`) in the Slack channel dedicated to the user’s session.
2. This command triggers the Back to Chatbot action.
3. The MxChat plugin updates the chat session, switching the user back to the AI chatbot.
4. The user receives a message in the chat interface indicating they are now interacting with the AI chatbot.

Product Recommendations

Location: MxChat → Actions → WooCommerce Features Category

Badge: Pro

Description: The Product Recommendations action leverages AI to provide personalized product suggestions to users based on their conversation, order history, and cart contents. When triggered, the chatbot analyzes the user’s query and context, then generates a list of recommended products with brief justifications, enhancing the shopping experience and driving sales.

Requirements:

• WooCommerce Add-on: The MxChat WooCommerce add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.
• WooCommerce Store: A functioning WooCommerce store with products is necessary.
• OpenAI API Key: A valid OpenAI API key is required. This is the same API key used for the chat model.
• MxChat Configuration: The OpenAI API Key must be configured in the MxChat settings under MxChat → Chatbot Tab.

Configuration:

1. Ensure the MxChat WooCommerce Add-on is installed and activated.
2. Activate your MxChat Pro license.
3. Navigate to MxChat → Actions and either create a new action or edit an existing one.
4. In the action configuration, select Product Recommendations as the action type.
5. Enter a descriptive Action Label (e.g., “Product Recommendations”).
6. Define Trigger Phrases that will initiate the product recommendations. Examples include “recommend me some products”, “what should I buy”, “suggest some items”, etc. See example phrases below.
7. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
8. Assign the action to the desired Bots.

Example Trigger Phrases:

• Recommend me some products
• What should I buy?
• Suggest some items
• I need some recommendations
• What do you recommend?
• Help me find something to buy

Order History

Location: MxChat → Actions → WooCommerce Features Category

Badge: Pro

Description: The Order History action allows users to quickly access information about their past orders directly within the chat. When triggered, the chatbot retrieves the user’s order history and uses AI to provide a conversational summary, answer specific questions about orders, or guide them through the order details. This provides a convenient way for customers to track their purchases and get support without leaving the chat interface.

Requirements:

• WooCommerce Add-on: The MxChat WooCommerce add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.
• WooCommerce Store: A functioning WooCommerce store with products and orders is necessary.
• User Login: Users must be logged in to view their order history.
• OpenAI API Key: A valid OpenAI API key is required. This is the same API key used for the chat model.
• MxChat Configuration: The OpenAI API Key must be configured in the MxChat settings under MxChat → Chatbot Tab.

Configuration:

1. Ensure the MxChat WooCommerce Add-on is installed and activated.
2. Activate your MxChat Pro license.
3. Navigate to MxChat → Actions and either create a new action or edit an existing one.
4. In the action configuration, select Order History as the action type.
5. Enter a descriptive Action Label (e.g., “Order History”).
6. Define Trigger Phrases that will initiate the order history retrieval. Examples include “what are my orders”, “show me my order history”, “track my order”, etc. See example phrases below.
7. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
8. Assign the action to the desired Bots.

Example Trigger Phrases:

• What are my orders?
• Show me my order history
• Track my order
• What did I order?
• My past purchases
• Where is my order?

Show Product Card

Location: MxChat → Actions → WooCommerce Features Category

Badge: Pro

Description: The Show Product Card action displays a visually appealing product card directly within the chat interface, providing users with key product information, pricing (optional), and an “Add to Cart” button (optional). When a user inquires about a specific product or expresses interest in a category, this action displays a rich product card, allowing them to quickly learn more and make a purchase without leaving the chat window.

Requirements:

• WooCommerce Add-on: The MxChat WooCommerce add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.
• WooCommerce Store: A functioning WooCommerce store with products is necessary.
• OpenAI API Key: A valid OpenAI API key is required. This is the same API key used for the chat model.
• MxChat Configuration: The OpenAI API Key must be configured in the MxChat settings under MxChat → Chatbot Tab.

Configuration:

1. Ensure the MxChat WooCommerce Add-on is installed and activated.
2. Activate your MxChat Pro license.
3. Navigate to MxChat → Actions and either create a new action or edit an existing one.
4. In the action configuration, select Show Product Card as the action type.
5. Enter a descriptive Action Label (e.g., “Product Inquiry”).
6. Define Trigger Phrases that will initiate the product card display. Examples include “tell me about “, “what is “, “show me “, etc. See example phrases below.
7. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
8. Assign the action to the desired Bots.
9. Configure the following settings in MxChat → Toolbar & Components → Toolbar Settings:
   • Show Product Prices: Enable to display product prices on the product card.
   • Show “Add to Cart” Button: Enable to display the “Add to Cart” button on the product card.
   • “Add to Cart” Button Text: Customize the text on the “Add to Cart” button.

Example Trigger Phrases:

• Tell me about
• What is ?
• Show me
• I want to know about
• Describe
• What are the details of ?

Add to Cart

Location: MxChat → Actions → WooCommerce Features Category

Badge: Pro

Description: The Add to Cart action allows users to quickly add products to their WooCommerce cart directly from the chat interface. When a user expresses intent to purchase a product, the chatbot identifies the product and adds it to their cart, providing a seamless and convenient shopping experience. It can be triggered by natural language requests or by clicking an “Add to Cart” button on a product card.

Requirements:

• WooCommerce Add-on: The MxChat WooCommerce add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.
• WooCommerce Store: A functioning WooCommerce store with products is necessary.
• OpenAI API Key: A valid OpenAI API key is required. This is the same API key used for the chat model.
• MxChat Configuration: The OpenAI API Key must be configured in the MxChat settings under MxChat → Chatbot Tab.

Configuration:

1. Ensure the MxChat WooCommerce Add-on is installed and activated.
2. Activate your MxChat Pro license.
3. Navigate to MxChat → Actions and either create a new action or edit an existing one.
4. In the action configuration, select Add to Cart as the action type.
5. Enter a descriptive Action Label (e.g., “Add to Cart”).
6. Define Trigger Phrases that will initiate the add-to-cart action. Examples include “add to cart”, “I want to buy “, “put in my basket”, etc. See example phrases below.
7. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
8. Assign the action to the desired Bots.

Example Trigger Phrases:

• Add to cart
• I want to buy
• Put in my basket
• I’ll take
• I want to purchase
• Add this to my cart

Proceed to Checkout

Location: MxChat → Actions → WooCommerce Features Category

Badge: Pro

Description: The Proceed to Checkout action allows users to quickly and easily navigate to the WooCommerce checkout page directly from the chat interface. When triggered, the chatbot confirms the contents of their cart and then immediately redirects them to the checkout page, streamlining the purchase process.

Requirements:

• WooCommerce Add-on: The MxChat WooCommerce add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.
• WooCommerce Store: A functioning WooCommerce store with a configured checkout page is necessary.

Configuration:

1. Ensure the MxChat WooCommerce Add-on is installed and activated.
2. Activate your MxChat Pro license.
3. Navigate to MxChat → Actions and either create a new action or edit an existing one.
4. In the action configuration, select Proceed to Checkout as the action type.
5. Enter a descriptive Action Label (e.g., “Go to Checkout”).
6. Define Trigger Phrases that will initiate the checkout redirection. Examples include “go to checkout”, “checkout now”, “I’m ready to pay”, etc. See example phrases below.
7. Set the Similarity Threshold to determine how closely the user’s query must match the trigger phrases. Adjust this based on the desired sensitivity.
8. Assign the action to the desired Bots.

Example Trigger Phrases:

• Go to checkout
• Checkout now
• I’m ready to pay
• Proceed to checkout
• Let’s checkout
• Take me to checkout

Perplexity Research

Location: MxChat → Actions → Search Features Category

Badge: Pro

Description: The Perplexity Research action allows the chatbot to conduct in-depth research using the Perplexity AI search engine and present a summarized answer to the user. This provides users with quick access to reliable information and insightful summaries without leaving the chat interface.

Requirements:

• Perplexity Add-on: The MxChat Perplexity add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.
• Perplexity API Key: A valid Perplexity API key is required and configured in the Perplexity Add-on settings.

Additional Information:

This action is more complex in how it’s originated. You will be able to find out more information on it in the Perplexity Research add-on documentation section.

Form Collection

Location: MxChat → Actions → Form Features Category

Badge: Pro

Description: The Form Collection action enables the chatbot to collect structured information from users by presenting them with interactive forms. This allows you to gather leads, collect feedback, or gather any other type of data in a conversational manner.

Requirements:

• Forms Add-on: The MxChat Forms add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.

Additional Information:

This action is more complex in how it’s originated. You will be able to find out more information on it in the Form Collection add-on documentation section.

Special Notes: Form Collection actions are often automatically created by the Forms add-on. Use the “Edit Form” button in the Forms add-on to manage the associated form.

Smart Recommender Flow

Location: MxChat → Actions → Recommendation Features Category

Badge: Pro

Description: The Smart Recommender Flow action allows you to guide users through a series of questions to provide personalized recommendations based on their preferences. This action is powered by a sophisticated flow builder, enabling you to create complex recommendation logic.

Requirements:

• Smart Recommender Add-on: The MxChat Smart Recommender add-on must be installed and activated.
• Pro License: An active MxChat Pro license is required to use this feature.

Additional Information:

This action is more complex in how it’s originated. You will be able to find out more information on it in the Smart Recommender Flow add-on documentation section.

Special Notes: Smart Recommender Flow actions are often automatically created by the Smart Recommender add-on. Use the “Manage Flows” button in the Smart Recommender add-on to manage the associated recommendation flow.