Microsoft introduced agents (or "Copilot agents") as part of its evolving AI strategy in 2024. The announcement was made at Microsoft Build 2024. Microsoft agents are advanced AI assistants that can automate complex tasks and run autonomously across multiple platforms, including Microsoft 365 (Word, Excel, Teams), Power Platform (Power Automate, Power Apps), and Azure AI. These agents go beyond traditional copilots by being able to proactively trigger actions, monitor processes, and integrate deeply into enterprise systems to manage workflows.
This change emphasize the Microsoft Copilot strategy that I spoke about previously this year. In this article, we will focus on Copilot Agents and Autonomous Agents.
What is a copilot agent ?
An agent is an artificial intelligence entity capable of performing tasks autonomously by interacting with systems, users, and data. Unlike traditional virtual assistants (such as chatbots, or virtual agents), agents are designed to make decisions, adapt to varied contexts, and execute actions proactively without requiring continuous instructions from the user. They are autonomous, and have the ability to execute complex workflows leveraging automations with tools like Power Automate or Azure Logic Apps.
1. Copilot Agents
As expected, Microsoft is once again bringing AI to all users. This time, directly through Office 365, everyone will get a taste of the citizen developer experience. By creating an agent, users can leverage Copilot Studio's features to develop a more specialized and efficient copilot tailored to specific domains. There are many sources or knowledges to ground from :
- Documents – Upload static files such as PDFs, Word documents, and presentations to create a reliable and easily accessible knowledge base.
- SharePoint – Leverage SharePoint sites to aggregate and organize valuable information, making it a seamless and collaborative starting point for knowledge management.
- Dataverse – Dataverse allows you to centralize and securely manage structured data, making it an excellent source for dynamic and relational knowledge.
- Website – Websites can be indexed to extract valuable content and insights, transforming them into a rich knowledge base.
- Other business applications – Applications like Azure SQL or Salesforce can also serve as valuable knowledge sources.
With the agents available in Copilot 365, employees have the ability to create and deploy multiple specialized agents, each designed to meet specific needs in different areas. These agents are able to handle complex tasks in a targeted manner, thus offering in-depth expertise and superior performance compared to the general Copilot.
Agents are not efficient if they have several different responsibilities, make them expert in a specific domain. By relying on these specialized agents, users benefit from more relevant and efficient assistance, adapted to the specificities of their business or project.
2. Autonomous Agents (Copilot Studio)
Autonomous Agents, as the name suggests, bring greater proactivity to our workflows. In my view, this is where the true revolution is happening. An autonomous agent in Microsoft Copilot Studio is an artificial intelligence entity that can perform tasks autonomously, triggering actions and making decisions without direct human intervention. Unlike traditional copilots that assist the user in real time, these agents are designed to work in the background, monitoring events and executing workflows proactively.
You will no longer be limited to creating simple chat experiences; instead, you can extend across a wide range of triggers and automate complex workflows. For example, an agent can be activated when an email is received, a record is updated in Dataverse, or a task reaches a critical stage, enabling seamless and intelligent process management. This is also called Agentic AI which is a trend that will impact the industry for the coming years (if not already).
Conclusion
In conclusion, the integration of Copilot agents and autonomous agents in Microsoft 365 revolutionizes the way companies automate and optimize their processes. These specialized agents allow users to create tailor-made solutions, adapted to specific areas, thus offering increased efficiency compared to general copilots. Thanks to Copilot Studio, organizations can now develop agents capable of triggering actions based on various events, such as the receipt of an email or the update of a record in Dataverse, facilitating intelligent and proactive management of workflows. This technological advancement marks a significant step towards deeper automation, freeing employees from repetitive tasks and allowing them to focus on higher value-added activities.