Skip to content

Advanced Triggers

In this tutorial, you'll explore more advanced trigger patterns and use cases in Playbooks AI.

Objective

By the end of this tutorial, you'll understand:

  • How to create complex data validation triggers
  • How to use sentiment and intent-based triggers
  • How to implement communication-based triggers
  • How to build robust error handling with triggers

Prerequisites

Data Validation Triggers

One powerful application of triggers is data validation. You can create playbooks that automatically trigger when users provide certain types of data:

## Main
### Triggers
- At the beginning
### Steps
- Tell the user this is a banking system
- Ask the user for their account number
- Ask the user for their PIN
- Load account information
- ...

## AccountNumberValidation
### Triggers
- When user provides an account number
### Steps
- While account number is not exactly 10 digits
  - Tell the user their account number is invalid
  - Ask the user to provide a valid 10-digit account number
  - If the user wants to quit
    - End program
- Return the account number

## PinValidation
### Triggers
- When user provides a PIN
### Steps
- While PIN is not exactly 4 digits
  - Tell the user their PIN is invalid
  - Ask the user to provide a valid 4-digit PIN
  - If the user has made 3 failed attempts
    - Lock user account
    - Tell the user their account is locked
    - End program
- Return the PIN

This pattern:

  • Separates validation logic from the main workflow
  • Automatically triggers validation when specific data is provided
  • Provides clear error messages and recovery paths

Sentiment-Based Triggers

You can create triggers that respond to user sentiment, enabling more natural conversations:

## AngryCustomerResponse
### Triggers
- When user is frustrated or abusive
### Steps
- Apologize to the user for the frustration
- Offer to connect them with a human support agent
- Ask if they would like to continue or speak with a human
- ...

This trigger activates when the system detects negative sentiment in the user's messages, allowing for empathetic responses at the right time.

Intent-Based Triggers

Similar to sentiment, you can create triggers based on user intent:

## AccountBalanceInquiry
### Triggers
- When user wants to know their account balance
### Steps
- ...

This trigger activates when the user expresses a need for help or shows signs of confusion, providing assistance without explicit commands.

Communication-Based Triggers

In multi-agent systems, triggers can respond to messages from other agents:

## StockUpdateNotification
### Triggers
- When you receive message from InventoryAgent asking for stock level
### Steps
- ...

This allows agents to communicate and react to information shared by other agents.

Error Handling with Triggers

Triggers can implement robust error handling patterns:

## MissingPlaybook
### Triggers
- When no suitable playbook is available
### Steps
- Tell the user that an error occurred
- ...

This creates a centralized error handling approach that can cover your entire application.

Combining Multiple Trigger Types

For complex scenarios, you can combine multiple trigger types:

## SecurityAlert
### Triggers
- When user attempts more than 3 failed logins
- When user attempts more than 3 failed credit card transactions
### Steps
- Log the security concern
- ...

This pattern creates sophisticated security monitoring by evaluating multiple risk factors.

Best Practices for Advanced Triggers

  • Prioritize user experience: Use triggers to make conversations feel natural and responsive
  • Handle edge cases: Create specific triggers for error conditions and unusual scenarios
  • Test thoroughly: Complex trigger systems need careful testing with various user inputs
  • Document your trigger strategy: Make it clear which triggers exist and when they activate

Exercises

  1. Implement a simple customer support agent who helps with order status and account renewal requests.

Next Steps

Now that you understand advanced trigger patterns, you're ready to explore Multi-Agent Programming to create systems with multiple specialized agents.