7 October 2026
Microsoft is making one of its biggest bets yet on AI agents, transforming Windows from an operating system that runs applications into a platform where autonomous AI systems can perform work.
The company announced new developer tools today that allow AI agents to execute tasks directly on Windows computers while being restricted by security policies.
The move puts Microsoft directly into competition with Apple and the major AI companies over what could become the next major computing interface.
The next computer interface may be an AI agent
For decades, using a computer meant opening applications, clicking menus and typing commands.
AI agents could change that.
Instead of opening a spreadsheet and manually analysing data, a user could tell an AI agent what needs to be done.
Instead of opening a development environment, the agent could write and test code.
Instead of searching through folders, the agent could locate the information required for a task.
Microsoft wants Windows to become the environment where those agents operate.
Security is now the biggest problem
Giving AI control over a computer is considerably more complicated than giving it access to a chatbot window.
An agent may need to read files, use applications and execute commands.
If it makes a mistake, the consequences can be serious.
Microsoft therefore announced Execution Containers, designed to create controlled environments in which agents can operate without automatically gaining unrestricted access to the entire computer.
OpenAI and Anthropic are part of the strategy
Microsoft says OpenAI, Anthropic and Nvidia are among the companies using the new technology.
That is significant.
Microsoft is not simply building an AI assistant for Windows.
It is trying to create an underlying platform that multiple AI companies can use.
That could make Windows an important operating environment for the agentic-AI era.
Europe has an additional reason to care
European businesses are particularly interested in controlling what AI agents can access.
An employee might eventually have several agents working on their computer.
One could manage documents.
Another could write code.
Another could analyse customer information.
Another could automate administrative tasks.
Companies will need to determine exactly what those agents are allowed to see and change.
Microsoft’s new controls are therefore not simply a technical feature.
They could become part of the basic security infrastructure required for corporate AI adoption.
The PC is becoming an AI computer
Microsoft’s new Surface hardware makes the strategy even clearer.
Powerful local processors and AI accelerators can allow more AI processing to happen directly on the machine.
The cloud remains important, but the PC itself becomes part of the AI infrastructure.
That could eventually lead to computers that are constantly running specialised AI agents in the background.
The important question is no longer whether a PC can run AI.
It is becoming:
How much work can an AI agent safely do on your computer without you having to do it yourself?
Microsoft is betting that the answer will be: quite a lot.
Laat een reactie achter