Day 24 - Understand Copilot Scopes, Skills, and Mitigation Layers
In today's video we'll bring it all together in a way that can help you understand the variance in responses you get when you work in ChatGPT, versus a PowerPoint or Power Apps Copilot, versus a chatbot from Copilot Studio. What makes the difference? What protective measures are implemented? Most importantly - what are the factors that help YOU decide what to use and when to make your own copilot!?
Side Note: Correction: The term "GPT powered" really refers to the ability to generate responses from natural language (available in all cases)....but that's not about the data sources they access (which you may think from the way I say that in the video). However, you'll see that the 'internet' is not a data source for most product copilots - as they are scoped to the task at hand and instead use Microsoft Graph rather than internet data (aka: Teams, PowerPoint, Excel, Power Apps built-in Copilots). However, in Copilot Studio you can mash up approaches to leverage many data sources, and further filter or constrain reponses.
Concept question: Where do you see copilots going in the next 5 years? next 10 years? Where do you want to see it go? Post your answers on our community page:
https://aka.ms/25DaysofCopilots
Full video playlist includes EVERYONE's videos plus extra Microsoft Videos:
• 25 Days of Copilots
Happy Holidays,
Audrie