Top Windows 11 exec tries to put out flames around agentic AI controversy (www.techradar.com)

Microsoft’s head of Windows, Pavan Davuluri, tried to tamp down a backlash after an earlier post suggesting Windows is “evolving into an agentic OS” sparked criticism. Davuluri rebooted the conversation on X, saying the Windows team “takes in a ton of feedback” and acknowledging concrete UX gaps — from inconsistent dialogs to power‑user workflows — and pledging to make Windows more reliable and developer‑friendly. His reply, however, skirted the louder complaints about promo nudges, telemetry and perceived ecosystem pushiness that many users and developers say undermine trust in a paid OS. The row crystallized around practical failures of the agentic Copilot demo and the broader Windows 11 update cadence: a clumsy Copilot walkthrough that misdirected users and suggested wrong settings, plus ongoing instability tied to rapid feature drip (notably since Windows 11 24H2 and the Germanium platform for Arm Copilot+ PCs). For the AI/ML community this episode matters because it highlights real-world limits of deploying agentic assistants at OS scale — the need for robust reliability, clearer UX boundaries, better developer tooling/APIs, and thoughtful rollout/testing. If Microsoft wants developers and enterprise users to buy into an AI‑driven desktop, it must fix fundamentals, curb promotional friction, and demonstrate trustworthy, predictable agent behavior.