Microsoft has decided to redesign the control of one of its most strategic bets in artificial intelligence. The company announced an internal reorganization to unify efforts of Copilot in consumer and enterprise divisions under a single structure, while reinforcing Mustafa Suleyman’s focus on advanced models and his mission of “superintelligence.” The change was communicated by Satya Nadella and Suleyman himself in separate messages to employees published by Microsoft.
The key takeaway isn’t just in the names, but in the direction they set. Microsoft asserts that AI is entering a new phase where it no longer simply answers questions or suggests code, but begins to execute multi-step tasks, with user control and real connections between agents, applications, and workflows. In this context, Nadella explains that Copilot should evolve from “a collection of great products” into “a truly integrated system,” built around four pillars: Copilot experience, Copilot platform, Microsoft 365 applications, and AI models.
This repositioning comes right after several clear signals about Microsoft’s future direction. In recent weeks, the company introduced Copilot Tasks as an evolution “from answers to actions,” meaning moving from providing responses to enabling action, and Copilot Cowork as a new automation layer within Microsoft 365 capable of streamlining tasks, coordinating workflows, and working with business context. Microsoft also positioned Work IQ as the intelligence layer giving context to Copilot through signals from Outlook, Teams, Excel, files, meetings, and organizational data.
Jacob Andreou takes charge of experience; Mustafa Suleyman focuses on models
The most visible part of this shift is the promotion of Jacob Andreou, who now leads the Copilot experience in both consumer and commercial segments as EVP, Copilot, reporting directly to Satya Nadella. According to Microsoft, Andreou will oversee design, product, growth, and engineering of that unified experience. Until now, he was responsible for product and growth at Microsoft AI.
Meanwhile, Mustafa Suleyman will continue reporting to Nadella but with an even more focused mission on the foundational layer: building impactful models in product, reducing operational AI costs, and advancing capabilities deemed critical for the next decade. In his internal message, Suleyman states that the company already has a roadmap for frontier-scale computing, and now the priority is to reorganize to dedicate all resources to creating cutting-edge models for Microsoft over the next five years.
The interesting aspect here is that Microsoft isn’t framing this move as just a corporate reorganization, but as a necessary adjustment to compete on two fronts: the models and the products where those models are integrated. Suleyman clearly articulates that the future will be defined by these two elements, and that the new structure aims to keep them more closely aligned and coordinated.
From assistant to operational work layer
For a tech-focused outlet, perhaps the most relevant part is this: Microsoft wants Copilot to move away from being perceived as a fragmented assistant spread across Word, Excel, Outlook, Windows, or Chat, and instead become a common operational layer across personal and enterprise productivity. This ambition is more evident in Microsoft 365 Copilot’s upcoming Wave 3, where the company discusses embedded agent capabilities within applications and an environment designed to delegate, monitor, and govern AI actions at scale.
Similarly, Microsoft has recently introduced what it calls a Frontier Worker Suite, integrated with Agent 365 and supported by Work IQ, with general availability expected in May 2026 according to Microsoft 365’s public roadmap. This reinforces the idea that the company aims to go beyond a corporate chatbot and push toward a platform where AI acts as a coordinator of digital work.
A defensive and offensive move simultaneously
The reorganization can also be seen as a response to competitive pressures. Google is strengthening Gemini in productivity and enterprise, OpenAI seeks more direct end-user engagement, and Anthropic is gaining weight in complex tasks and agents. Microsoft, on its part, seems to have accepted that simply integrating AI into many applications is no longer enough; it now needs to ensure that all of this is perceived as a coherent architecture, with a shared narrative and less friction among consumer, enterprise, models, and platform. This interpretation aligns with Nadella and Suleyman’s messaging about the need to unify Copilot and align product, infrastructure, and models under a common leadership.
It remains to be seen whether this internal redesign will reduce overlaps and speed up the delivery of clearer experiences for users. But the strategic message is already clear: Microsoft does not want to manage Copilot as a collection of separate initiatives. It aims to make it a singular, cross-cutting, more aggressive system, while doubling down on the models layer that should support this leap. In a market where AI is increasingly measured by actual execution rather than demos, this difference could be decisive.
Source: Noticias Inteligencia Artificial