What Microsoft Is Really Building With AI

About this video

Kevin Stratvert reports back from Microsoft Ignite in San Francisco, which he attended in November — disclosing that Microsoft covered his travel but did not review or approve the video. Beyond keynotes, he sat down with the Windows and Edge teams and attended an Ask Me Anything session with Microsoft’s responsible AI leaders. His reading is that Ignite was not mainly about new AI features, but about a shift from AI that answers questions to AI agents that take action and quietly run routine work in the background, embedded in tools people already use. Copilot, he argues, is the front door/interface, while agents do the work behind the scenes across Windows, the browser, Microsoft 365 apps and Azure. In conversation with Pavan Davuluri, who leads Windows and devices, Windows is positioned as the secure, trusted foundation for AI, using a hybrid approach with some AI in the cloud and some running on-device for speed, privacy and awareness of what is on screen, adding voice and vision interaction. In conversation with Sean Lindersay, who leads Microsoft Edge, the browser is reframed from a page viewer into an agent that can visit multiple sites in background tabs and complete tasks for you, designed for secure enterprise work. A recurring theme was responsible AI: because a wrong action in a business system is worse than a wrong answer, Microsoft emphasised observability, testing and governance — seeing what AI is doing and controlling how it is used — as what makes AI safe enough to deploy at scale.

Key takeaways

• Microsoft’s Ignite strategy, as presented, is a shift from AI that answers questions to agents that take actions and run routine work in the background
• Copilot is framed as the interface/front door; the actual work is done by agents embedded across Windows, Edge, Microsoft 365 apps and Azure
• Windows is positioned as the secure foundation, mixing cloud AI with on-device AI for faster response, privacy and screen context, plus voice and vision input
• Edge is being reframed from a web viewer into an agent that can visit sites and complete tasks on the user’s behalf, with enterprise security as the differentiator
• Responsible AI — observability, testing and governance — is presented as the precondition for enterprises trusting agents that can act, not just answer

Episode details: published December 22, 2025 · 8:04 long. Summary written by Self Growth Videos from the episode itself. Business results and any income figures mentioned are Kevin Stratvert’s own claims, shared for education — they are not typical or guaranteed.

← Previous: BEST Free AI Search Tools of 2025 (ChatGPT Isn’t Enough) · Next: Claude Code Tutorial for Beginners →

All Kevin Stratvert videos · Kevin Stratvert profile · Kevin Stratvert on YouTube