Skip to content
My Docs

Create Power Platform solutions with AI and Copilot

Source: Create Power Platform solutions with AI and Copilot - Training | Microsoft Learn and my personal notes from work session Getting Started with Copilot Studio in 2024

Getting Started with Copilot Studio 2024-02-09

Section titled “Getting Started with Copilot Studio 2024-02-09”

Presented by Sravani Seethi, Microsoft Cloud Solution Architect

Use case: quickly create custom chatbots to help users

Functionality:

  • Conversational AI
  • Monitoring
    • Telemetry
    • User analytics, satisfaction
  • Publish copilot to different SaaS
    • Like Teams
  • Chat on knowledge from files, websites, shares
  • Create specific topic
    • Predefined responses
    • Custom responses
  • Actions and Plugins - use Power Platform, OpenAI connectors to call backend and APIs
    • Help users change data, query data
      • Can use Power Automate flow to change data for user

Demo and Review of Copilot Studio Interface

Section titled “Demo and Review of Copilot Studio Interface”

  • Access Copilot Studio from Power Platform - Chatbots

  • Go to https://web.powerva.microsoft.com/

  • Copilots

    • Plugins
    • Prompts
      • Reusable across Power Platform
      • Give GPT model the context of the conversation and input to give a relevant response
        • For example: I am a nurse at a hospital and need concise instruction on how to use a <medical device>. Do not add additional text.
    • Models

  • Create new copilot, feedback it website

    • Topics - there are built in ones, system defined like escalate to human

  • Create a new topic, can use copilot to create topic

    • Topic can use inputs, AI will understand variations of the inputs

  • Test copilot

  • Topic actions:

    • Send messages
    • Ask questions
    • Ask an adaptive card (HTML output)
    • Conditions
    • Variables
    • Manage topic like start new or end
    • Call an action like Power Automate flow
    • Other advanced features like generative answers, custom events, connector actions

  • Entities: outlines ways an user can respond in chat like different date formats and statements

  • Analytics

  • Publish

    • Different channels: Teams, Power Platform, Slack, Website

  • Extend:

    • Plugin

  • Settings

    • Add other websites, documents to feed
    • Content moderation
    • Generative AI
      • Can use public websites and internal SharePoint sites
    • Customer Engagement - like live agent integration
    • Security - sharing, authenticate user
      • No authentication allow all web channels publishing
    • Skill - for example Azure bot framework
    • Language - like English, French

Where is data stored?

  • Azure OpenAI does not use your data to train the model and data lives in your tenant