What are AI Actions?

AI Actions are powerful capabilities that extend your chatbot's functionality beyond simple conversations. They allow your chatbot to perform specific tasks automatically during conversations, such as collecting leads, searching products, escalating to live chat, and more.

Key Features

• Pre-built Actions: Most AI actions come with predefined configurations that you can simply enable or disable
• Smart Triggering: The AI automatically determines when to use each action based on conversation context
• Customizable Behavior: Configure when and how each action should be triggered
• Category Organization: Filter and browse actions by specific groups or categories using the sidebar

How to Enable AI Actions

Step 1: Access AI Actions Settings

1. Navigate to your bot's settings in the admin dashboard
2. Click on the AI Actions section in the sidebar
3. Browse available actions or use category filters to find specific types of actions

Step 2: Enable/Disable Actions

• Toggle the switch next to any AI action to enable or disable it
• Most actions work immediately with their default configuration
• Custom API integrations require additional configuration (see below)

Configuring AI Actions

Each AI action has three key configuration elements:

1. Trigger Prompt (Behavior Description)

Purpose: Describes the circumstances when this AI action should be triggered

What to include:

• Clear description of the situation or user intent that should activate this action
• Conversation context that indicates the action is needed
• User phrases or questions that signal this action should be used

Example:

"When the user wants to know about product availability, pricing, or specifications use getProductDetails action. Also use when user asks 'do you have...', 'how much is...', or 'tell me about...'"

For pre-built actions: This is already configured with optimal defaults

For custom API integration actions: You must clearly state when the AI should call your custom API

2. Action Label

Purpose: Text displayed in the chat interface while the chatbot is performing this action

Best practices:

• Keep it short and user-friendly
• Use present continuous tense (e.g., "Searching products...", "Collecting information...", "Connecting to agent...")
• Let users know what's happening without technical jargon

Examples:

• "Searching our product catalog..."
• "Submitting your request..."
• "Connecting you to a live agent..."
• "Checking availability..."

3. AI Action Definition

Purpose: Defines how this function works and what it does technically. It helps the model understand the tool’s purpose so it can call it correctly. A clear, detailed description ensures accurate tool usage, while vague ones lead to missed or incorrect calls.

What it includes:

• Technical implementation details
• Function parameters and expected inputs
• Return values and outputs

For pre-built actions: This is fully configured and managed by the system

For custom API actions: You must provide clear function description. Always refer to parameter which needs to be provided by chatbot or user to call this AI action.

Action Categories

AI Actions are organized into categories for easy navigation:

• Lead Generation: Collect user information, emails, and contact details
• E-commerce: Product search, catalog browsing, order tracking
• Live Chat: Escalation to human agents, support handoff
• Custom Integrations: Your own API endpoints and custom functionality
• Utility: General-purpose actions like scheduling, calculations, etc.

Use the category sidebar to filter and focus on relevant actions for your use case.

Best Practices

When Configuring Trigger Prompts

1. Be Specific: Clearly describe the exact scenario when the action should trigger
2. Include Examples: Provide sample user phrases that indicate the action is needed
3. Avoid Overlap: Ensure different actions have distinct trigger conditions
4. Test Thoroughly: Try various conversation flows to verify correct triggering

When Setting Action Labels

1. User-Friendly Language: Avoid technical terms
2. Progress Indication: Let users know something is happening
3. Consistency: Use similar patterns across all your actions
4. Brief: Keep labels under 50 characters

General Tips

• Start with pre-built actions before creating custom ones
• Enable only the actions your bot actually needs
• Test each action in a live chat to ensure proper behavior
• Update trigger prompts based on actual conversation patterns

Custom API Integration Actions

For advanced users who want to integrate their own APIs:

Requirements

1. When to use this action - trigger prompt: You must clearly describe when to call your API - refer to its internal name - AI Action internal name
2. Action Label: Provide user-friendly status text
3. AI Action Definition: Complete technical specification including function description and parameter names with their types. Do not include information about integration details (such as URL, HTTP method type, headers etc). Only refer to parameters which needs to be provided by user or chatbot as a result of ongoing conversation or actions called already.

Troubleshooting

Action Not Triggering

• Review your trigger prompt for clarity
• Test with exact phrases from your trigger description
• Check if another action might be triggering instead
• Verify the action is enabled

Action Triggering Too Often

• Make trigger prompt more specific
• Add exclusion conditions
• Review conversation logs to identify false positives

Custom API Not Working

• Verify API endpoint is accessible
• Check authentication credentials
• Review API definition format
• Test API independently to ensure it's responding correctly