From the course: Copilot Agents: Build Your Own AI Assistant in Copilot and Teams (No Code Required) (February 2025)
Explore next steps with Copilot Studio - Microsoft Copilot Tutorial
From the course: Copilot Agents: Build Your Own AI Assistant in Copilot and Teams (No Code Required) (February 2025)
Explore next steps with Copilot Studio
- You may have noticed that there are some limitations to the agents that you can make using the agent builder. If you reach the limit of what you can do here, Microsoft does offer a separate tool called Copilot Studio, which has a much more robust set of features. Now, this workshop does not teach how to use the full version of Copilot Studio, but I did want to give a brief introduction. Copilot Studio requires another separate purchase. Your organization's Microsoft Administrator can purchase and enable it for your team. Instead of paying a subscription price per user, Copilot Studio will be enabled for the entire organization and your administrator will have to pay for the number of messages processed by your agents. There's some information for administrators looking to activate Copilot Studio here, but let's go back to Copilot where we've been working, and when I click the option to create an agent, you may have noticed that this page is labeled as Copilot Studio, and this might be a little misleading. We have not been using Copilot Studio here. What we've been using is technically called Copilot Studio Agent Builder, and it's just a light version of the tool set that's based on Copilot Studio. If your organization does have access to it, you can go to copilotstudio.microsoft.com. In Copilot, you can create an agent, and there's also a library of pre-made templates you can use as a starting point, or you can make a new basic agent starting with tools that are similar to what we've used in this workshop. You can describe the parameters in the chat interface or switch to the manual configure mode, but the tool set really expands when you open an agent that you've already created. You can adjust the agent's instructions and knowledge sources, but you can also include actions, which give the agent the ability to complete tasks in other applications or online services. You can include triggers, allowing the agent to automatically take action when certain events occur, and topics let you create established conversation paths, so the agent will have specific responses to specific user inputs, or you can enable generative AI orchestration, which gives the agent the ability to choose some of its own actions, knowledge sources, and topics. And when you're finished, Copilot Studio can publish your agents so they can be used by your teammates in Copilot and Teams, but it also has the ability to publish agents for external customers and partners, or it can format the agent as a chat bot that you can place on a website or publish on third-party communication platforms. So Copilot Studio offers a much more comprehensive and complex set of tools for creating autonomous agents. For some people, it offers valuable next steps to look into when you reach the limit of the agent builder that we've learned in this workshop.