Windows 11: Microsoft uses AI to detect the causes of PC slowdowns

Windows 11: Microsoft uses AI to detect causes of PC slowdowns

After implementing numerous integrations of Copilot into Windows 11, Microsoft is now exploring a much more practical use of artificial intelligence. This time, the AI is not focused on composing text or summarizing documents, but rather on assisting engineers in understanding why a PC is slowing down.

The company is currently testing an AI integration in the Windows Performance Analyzer (WPA), a tool used internally by Microsoft, hardware manufacturers, and developers to analyze system performance. This development could significantly speed up the diagnosis of complex issues.

An AI Capable of Interpreting Performance Data

Windows Performance Analyzer is one of the most powerful tools in the Windows ecosystem. It allows for the examination of thousands of system events to identify the cause of excessive resource consumption, slowdowns, or abnormal behavior.

Until now, this analysis required substantial expertise. Engineers had to manually sift through numerous charts, tables, and system logs to piece together the sequence of events.

With this new integration, the approach is completely transformed.

Users will be able to query the tool in natural language. Questions like “What caused this CPU usage spike?”, “Is the disk responsible for the slowdown?” or “Which process is creating the bottleneck?” will enable the AI to automatically analyze performance traces and produce a summary with relevant technical details.

Copilot Relies on the MCP Protocol

This feature is based on the Model Context Protocol (MCP), a protocol designed to enable AI assistants, like GitHub Copilot, to interact directly with specialized tools.

In this scenario, Copilot does not replace Windows Performance Analyzer. Instead, it acts as an interpretative layer that connects the charts, statistics, and tables of the application to understandable explanations.

The goal is to reduce the time needed to identify anomalies while facilitating the processing of a large volume of technical data.

Microsoft emphasizes that performance analysis currently relies on very specialized skills. Even experienced users may spend several hours correlating different views before isolating the true cause of a problem.

A Technology Primarily Designed for Engineers

Unlike some Copilot features aimed at the general public, this development is primarily oriented towards developers, PC manufacturers, and teams responsible for Windows development.

Thus, it is not a tool that will automatically fix a slow computer.

Microsoft is concurrently developing PC Insights, a more accessible feature integrated into Copilot capable of explaining to a regular user the status of their CPU, RAM, storage, or battery.

The WPA integration operates at a completely different level, providing much deeper analyses intended for professionals.

A Technology Still in Experimental Phase

This new function, however, remains in a pre-release stage.

According to initial information, it currently requires GitHub Copilot and an active subscription. Microsoft also reminds users that the responses generated by the AI may be incomplete or contain errors.

Engineers will therefore need to verify the AI’s conclusions by consulting the raw data from system traces.

This approach reflects a growing trend in professional tools: AI is not intended to replace human expertise, but to accelerate analysis and reduce the time needed to reach a reliable diagnosis.

One of the Most Relevant Uses of Copilot

Since Copilot’s arrival in Windows, Microsoft has at times faced criticism for adding numerous AI features without genuine added value for users.

This experimentation adopts a different logic.

By helping engineers quickly identify faulty drivers, resource-intensive applications, or performance issues before they impact millions of computers, Microsoft is finally leveraging AI where it can truly save time.

If this technology lives up to its promises, it could transform how Windows is tested, optimized, and debugged, well before users encounter any slowdown on their own machines.




Scroll to Top