Using Inputs on Topics in Copilot Studio

Jul 14, 2024

Using Inputs on Topics in Copilot Studio

Introduction

  • New feature in Copilot Studio: inputs on topics
  • Similar concept to inputs and outputs on actions
  • Eliminates the need for question nodes in custom topics
  • Enables using an LLM for entity extraction and updating previous inputs seamlessly
  • Example: Ordering a pizza and changing the details mid-conversation

Benefits of Inputs on Topics

  • Simplifies creation of powerful custom topics with minimal effort
  • Supports hierarchical entities in slot filling
  • Contextual conversation flow with dynamic entity extraction
  • Streamlined user experiences

Demonstration Overview

  • Scenario: Scheduling a consultation with a business expert
  • Old Approach: Series of question nodes for collecting inputs
  • New Approach: Using inputs and outputs directly within topics

Practical Example

  1. Environment: Inside Copilot Studio's new UX
  2. Example Topic: Requesting a device repair
    • Trigger Phrase: "Request device repair"
    • Inputs: Model number and serial number
  3. Input Mechanism: Explanation
    • Provides variables without question nodes
    • Allows to change inputs dynamically in the conversation

Features of Inputs

  • Inputs and outputs can be defined within topic details
    • Includes variables like Surface model and serial number
    • Ability to create new variables as needed
  • Similar to action inputs
    • Includes entity validation and other question node capabilities

Outputs

  • Outputs can also be defined with descriptions and examples
  • Enhances bot's interaction effectiveness

Conclusion

  • Inputs on topics are transformative for building bots in Copilot Studio
  • Encourages dynamic, context-aware conversation design
  • Practical impact: Streamlining bot development and enhancing user experiences

Call to Action: Try Copilot Studio at aka.ms TR copilot Studio

  • Like and subscribe for more updates.