AI Agents in Microsoft Copilot
In an era where technological breakthroughs weave into the fabric of daily life, the concept of AI Agents in Microsoft’s environment could seem like something out of a science fiction reality. Yet, the tutorial video from October 9, 2025, by Shervin Shaffie, an engineer at Microsoft, brings this subject into practical focus, demonstrating how Microsoft 365’s Copilot platform enables users to build AI agents without extensive coding knowledge. The video, published by ‘Collaboration Simplified’ and currently boasting over 66,000 views, dives into pragmatic steps to create these digital assistants, effectively simplifying and automating tasks.
Shaffie begins by guiding viewers through logging into M365 Copilot, immediately immersing them in the ecosystem where these agents thrive. His practical approach and detailed walkthroughs demystify the process, showing how to use existing templates to create custom agents tailored to specific needs like the ‘Quiz Tutor’ or a ‘Position Pal.’ Such applications illustrate AI’s potential in performing repetitive tasks efficiently.
The engineer skillfully highlights the potential of Copilot Chat. The availability of templates enables users to focus on customizing only what they need, allowing for creativity without the risks of starting from scratch. Positive feedback emerges as Shaffie outlines the intuitive interface of Copilot Studio (Lite) that allows even the uninitiated to harness this powerful tool.
Yet, amidst the exciting innovations, Shervin addresses a significant limitation—the knowledge section’s restriction to public URLs. While acknowledging the lack of support for attachments like SharePoint files, he suggests workarounds that demonstrate profound adaptability and resourcefulness, such as “Ask me for the file,” as a way to circumvent these constraints effectively.
Examples of AI agents illustrate how business processes are streamlined. The video uses relatable scenarios, like the building of a ‘Position Pal’ agent for sorting and comparing resumes against job postings. This illustrates how AI can offload repetitive, detail-oriented tasks, allowing users to focus on more strategic activities.
Finally, the video touches on the expanded capabilities available to license holders, offering a glimpse into the potential power of advanced agents such as autonomous setups or multi-agent orchestration. Despite lacking these functionalities, the standard Copilot allows users a hands-on learning experience, cultivating a deep understanding of AI in everyday applications.
Shaffie’s tutorial is a masterclass in empowering users to engage with AI agents creatively and efficiently, even without advanced licensing. It reflects the wider trend of technology making complex AI tools accessible to a broader audience. By the video’s conclusion, viewers are left with an intriguing mix of practical tips and insights into potential innovations, sparking curiosity about what role these intelligent agents might play in our futures.