Chat Node

Overview #

The Chat Node lets you add an AI-powered chatbot anywhere on your WordPress site. Whether you want a standalone chatbot or one that’s connected to other parts of your workflow, the Chat Node gives you complete control over how your chatbot looks, behaves, and responds to users.

You can customize every aspect of your chatbot, from its appearance and position to the AI model it uses. You can even feed it your own data so it only uses that information to answer questions.

Key Features #

Basic Configuration #

• Node Name: Give your chat node an easy-to-remember name within your workflow
• System Prompt: Tell the chatbot how to behave, what to know, and how to respond
• AI Model Selection: Choose from many top AI models including:
  • OpenAI models (GPT-4o, o1, o3)
  • Claude models (Claude 3.7, 3.5, 3 Opus, etc.)
  • Gemini models
  • Llama models
  • Mistral models
  • DeepSeek models
  • Perplexity models
  • Grok models

Knowledge Customization #

• System Prompt: Define how your chatbot should respond and what information it has access to
• Node Connections: Include information from other nodes, such as:
  • User form inputs
  • Static content
  • API responses
  • Database queries

Automated Actions: Building AI Agents #

• Custom Actions: Transform your chatbot into a powerful AI agent that can perform complex tasks, such as:
  • Collecting contact information
  • Signing users up for newsletters
  • Booking appointments
  • Submitting support tickets
  • Processing orders
  • Generating personalized content
  • Triggering automation sequences
  • Providing customer support with real solutions (not just answers)
• Action Fields: Define what information each action needs to collect
• Action Connections: Connect each action to its own complete workflow, creating a network of specialized AI agents

Knowledge Enhancement Tools #

• Web Search: Let your chatbot search the internet for up-to-date information
  • Control search context size
  • Set geographic location for more relevant results
• File Search (RAG): Power your chatbot with your own knowledge using Retrieval-Augmented Generation
  • Connect to vector stores containing your documents
  • Include WordPress content (posts, pages) as knowledge sources
  • Set maximum number of search results
  • Show citations to maintain transparency

Design Customization #

• Theme Options:
  • Light theme
  • Dark theme
  • Custom theme with color pickers
• Position:
  • Bottom right (default floating widget)
  • Bottom left
  • Top right
  • Top left
  • Inline (embedded in content)
• Bot Identity:
  • Custom bot name
  • Icon selection (Robot, Assistant, Brain, Chat)
• Typography:
  • Font family selection (Inter, Arial, Georgia, Montserrat, Roboto)
  • Chat text size
  • Header text size
• Dimensions:
  • Custom width
  • Custom height
  • Border radius
• Quick Response Buttons:
  • Add pre-set buttons users can click
  • Customize button text and the message sent
  • Guide users through conversations

Behavior Settings #

• Initial Message: Welcome message shown when chat starts
• Input Placeholder: Custom placeholder text for the message input
• History Management:
  • Control how many messages to keep in history
  • Option to save chat history between visits
• Rate Limiting:
  • Prevent abuse by limiting messages
  • Set maximum messages per time window
  • Control time window duration
• Advanced Features:
  • Typing indicators to show the AI is responding
  • Sound effects for messages
  • Auto-open delay to start chats automatically
  • Show citations for web/file search results

Using the Chat Node #

Basic Setup #

1. Add the Node: Drag the Chat Node into your workflow
2. Name Your Node: Give it a recognizable name
3. Choose an AI Model: Select the model that fits your needs
4. Add a System Prompt: Tell the AI how to behave and what to know
5. Save Your Workflow: To get the shortcode for embedding

Setting Up Your System Prompt #

The system prompt is how you tell your chatbot what it should know and how it should act. Here’s how to write a good one:

To include data from other nodes in your workflow, use the input tags like this:

Creating AI Agents with Automated Actions #

The Automated Actions feature transforms your chatbot from a simple Q&A tool into a powerful AI agent system that can understand user needs and take meaningful action. Each action becomes a specialized capability that your AI agent can execute:

1. Click “Add Action” in the Actions tab
2. Name your action (e.g., “Newsletter Signup” or “Product Recommendation”)
3. Add a description of what the action does (helps your AI understand when to trigger this)
4. Add fields to collect information:
   • Field name (e.g., “Email” or “Budget”)
   • Field type (text, email, number, phone)
   • Mark if required
5. Save your action
6. Connect the action output handle to other nodes or complete workflows

What makes this so powerful is that each action can connect to an entire separate workflow. This means your chatbot becomes an intelligent front-end that can:

• Understand complex user requests: “I need to book a consultation for next week, preferably in the afternoon”
• Collect the right information: Automatically guide users through providing necessary details
• Trigger sophisticated processes: Each action can launch specific workflows that process the information, connect to external systems, update databases, or generate new content
• Provide real solutions: Not just answers to questions, but complete end-to-end solutions to user problems

For example, a single chatbot could:

1. Help users troubleshoot technical issues (connecting to a workflow that accesses your knowledge base)
2. Process product returns (connecting to your order management system)
3. Generate custom quotes (calculating pricing based on user requirements)
4. Book appointments (integrating with your calendar system)

All from a simple, conversational interface that feels natural to your users. The AI intelligently determines which action to trigger based on the conversation context.

When a user’s message indicates they need a specific service, the chatbot recognizes this intent and automatically guides them through the appropriate action flow, collecting all needed information in a conversational way.

Using RAG: Retrieval-Augmented Generation #

Retrieval-Augmented Generation (RAG) is a powerful feature that lets your chatbot find and use information from your own content. With RAG, your chatbot can:

1. Access your knowledge base: Use your own documents, posts, and pages as a source of truth
2. Provide accurate answers: Base responses on your specific information instead of generic knowledge
3. Stay up-to-date: Include your latest content and avoid hallucinations
4. Maintain your voice: Answer questions using information that matches your brand’s style and perspective

Setting Up RAG With Knowledge Bases #

The Chat Node includes a complete RAG system through its File Search tool:

1. Create a Knowledge Base:
   • Go to Settings → Knowledge Base Management
   • Click “Create Vector Store” to make a new knowledge base
   • Give it a name like “Company Documents” or “Product Information”
2. Add Content to Your Knowledge Base:
   • Upload files directly (PDF, Word, text files, etc.)
   • Import WordPress content (posts and pages)
   • Filter WordPress content by category, author, or date range
3. Connect the Knowledge Base to Your Chatbot:
   • In your Chat Node, go to the Model tab
   • Enable “File Search” under OpenAI Tools
   • Select your knowledge base from the dropdown
   • Set the maximum number of results to include
4. Customize the Citation Display:
   • In the Behavior tab, toggle “Show Citations” to control whether sources are visible to users

Web Search vs. RAG (File Search) #

Both tools enhance your chatbot’s knowledge, but they work differently:

Web Search:

• Searches the internet for current information
• Good for general questions and current events
• Can be configured with context size (low, medium, high)
• Optional location settings for geographically relevant results

File Search (RAG):

• Searches only your uploaded documents and WordPress content
• Perfect for company-specific information
• Ensures answers come from approved sources
• Shows citations to your actual content

Quick Response Buttons #

Quick response buttons give users pre-set options to click on:

1. Go to the Design tab
2. Open the “Quick Response Buttons” section
3. Click “Add Button”
4. Enter:
   • Button Text (what users see)
   • Message to Send (what gets sent when clicked)
5. Add as many buttons as needed

These buttons appear at the start of chats or after AI responses to guide users.

Embedding Your Chatbot #

After setting up your chatbot and saving your workflow, you’ll get a shortcode:

You can add this shortcode to:

• Pages
• Posts
• Sidebars
• Footers
• Any area that supports shortcodes

The chatbot will appear as configured – either as a floating widget in the corner or inline in your content. When you make changes to your Chat Node and save the workflow, your chatbot will update automatically without changing the shortcode.

Best Practices #

For Better Responses #

• Give clear instructions in your system prompt
• Connect relevant knowledge from other nodes
• Set boundaries for what the chatbot should and shouldn’t answer
• Test with different user questions before publishing

For Effective RAG Implementation #

• Organize your knowledge strategically: Create separate vector stores for different topics or departments
• Use high-quality content: Make sure your documents are well-written and contain accurate information
• Update regularly: Keep your knowledge base fresh with current information
• Test with specific questions: Try questions that should be answered from your knowledge base
• Balance completeness and performance: More documents means more knowledge but potentially slower response times
• Add context in your system prompt: Tell the AI how to use the retrieved information
• Show citations when appropriate: Enable citations for transparency when using company documents

For Better User Experience #

• Choose a position that doesn’t block important content
• Set appropriate rate limits to prevent abuse
• Use typing indicators to make conversations feel natural
• Add quick response buttons for common questions
• Customize the design to match your website

For Creating Powerful AI Agents #

• Think in terms of user problems: What complete solutions can your chatbot provide?
• Create specialized actions: Build actions for specific tasks your users commonly need
• Design end-to-end workflows: Connect each action to a workflow that delivers complete results
• Layer your AI capabilities: Start with simple actions and build up to more complex ones
• Combine with other nodes: Use data processing, API connections, and form nodes to create powerful backend processes
• Test user journeys: Walk through complete scenarios from chatbot to final outcome

For Security and Privacy #

• Don’t store sensitive information in the system prompt
• Be transparent about data storage in your privacy policy
• Set appropriate message history limits
• Consider disabling history persistence for sensitive topics

Troubleshooting #

Common issues and solutions:

Chatbot Not Loading:

• Verify your workflow is properly saved
• Check if the shortcode is formatted correctly
• Look for JavaScript errors in your browser console

Styling Issues:

• Check for theme compatibility
• Review custom CSS conflicts
• Test on mobile devices

Response Issues:

• Verify your system prompt is clear
• Check node connections
• Confirm AI model settings
• Check API keys and limits

RAG System Technical Details #

The built-in RAG system uses advanced technologies to make your content searchable:

• Vector Embeddings: Documents are converted into vector representations that capture their meaning
• Semantic Search: Queries find relevant content based on meaning, not just keywords
• Content Chunking: Documents are automatically split into manageable pieces for better retrieval
• Relevance Ranking: The most relevant information is presented first in responses
• Document Processing: Support for multiple file formats including PDF, DOCX, TXT, MD, CSV and JSON
• WordPress Integration: Direct import of posts and pages with filtering capabilities

The entire system is built with security and privacy in mind:

• Your content stays on your server
• Processing happens through secure API connections
• No training on your data – it’s only used for retrieval