Microsoft has plans to convert Windows 11 into a platform capable of operating artificial intelligence agents to perform tasks on behalf of users. This strategy aims to adapt computer usage to a new modality, combining locally processed data with cloud-hosted services.
The proposal was presented during an event focused on Surface and Windows devices and also primarily aims to attract both developers and content creators. However, the demonstrations generated questions about the practical benefits for the average user in their daily activities.
This functionality, called hybrid intelligence, allows for the combination of AI models installed directly on the computer with services available in the cloud. The logic behind this is to allow certain operations to be completed locally, which would help reduce costs associated with external processing.
During the event, Microsoft demonstrated the functioning of GitHub Copilot executing programming tasks on the device itself. Additionally, models that can run locally, such as OpenClaw, Hermes, and Perplexity agents, were shown, along with a deeper integration of Copilot into the Windows operating system.
CEO's Vision for Transformation
Satya Nadella, the company's CEO, emphasized the shift in perspective adopted by the company, commenting that 'I think today is the day we will start talking and thinking about the frontier ecosystem with perhaps slightly different contours.'
Despite the presentations, there was skepticism, as clear examples were lacking of how local models could assist people who do not work in AI systems development. For now, it seems Microsoft is focusing its efforts on capturing precisely this niche of specialized audience.
To entice professionals who need to run large models, the company introduced the Surface Laptop Ultra, which is equipped with Nvidia RTX Spark chips. This computer has the capacity to run models with over 120 billion parameters locally.
The product description states that it was developed for creators and developers who demand maximum performance and seek portability with a superior finish. However, the purchase price restricts access to the machine. The Surface Laptop Ultra starts at US$ 2,599 (equivalent to about R$ 13,515), and some models with RTX Spark can cost up to US$ 6,999 (approximately R$ 36,395).
Other brands, such as Asus, Dell, HP, Lenovo, and MSI, also offer equipment using this same platform. For those who work intensively with AI, the high investment may be justified by the significant reduction in cloud processing expenses; however, for the average consumer, justifying such a cost becomes more difficult without evident practical benefits.
An update that might be more relevant to the general public is the redesign of the Windows 11 Search function. Its interface has been redesigned to display results directly on the screen, functioning similarly to an application launcher.
Past Experiences and Future Challenges
Microsoft's previous trajectory with AI helps explain the current caution. The Copilot+ PCs, launched two years ago, highlighted NPUs, which are chips designed to accelerate artificial intelligence tasks. Despite this, few productivity, photography, or video tools have utilized these features so far.
The Recall feature, promoted as one of the highlights of the platform, lost importance due to privacy concerns raised and delays in its launch.
Computers equipped with RTX Spark are likely to have greater acceptance among developers and creators, as they have concrete reasons to run models locally. To reach a wider audience, it will be essential to make these capabilities accessible on lower-priced machines.
Beyond the hardware aspect, Microsoft faces a transformation that transcends physical equipment. Cloud-operated agents, utilizing Linux virtual machines, can increase compatibility with software from this operating system. The company needs to maintain its position in a scenario where the way computers are used is changing, which requires demonstrating that AI agents bring real advantages, and not just the need for more expensive computers for specialists.