On 17 March 2026, Microsoft announced a major reorganisation of its Copilot team. Mustafa Suleyman steps back from day-to-day product leadership to focus entirely on building proprietary frontier AI models. What does this mean for Microsoft's AI strategy and your organisation?
On 17 March 2026, Microsoft announced a major reorganisation of its Copilot organisation. The change goes beyond a personnel reshuffle: it marks a fundamental shift in the company's AI strategy.
Mustafa Suleyman, until now the public face of Copilot, steps back from day-to-day product leadership. He will now focus exclusively on developing new frontier AI models within Microsoft — including models for code generation, image and audio synthesis, and advanced reasoning.
Jacob Andreou, former senior executive at Snap, takes charge of the Copilot product experience for both commercial and consumer segments. The new Copilot Leadership Team consists of Suleyman, Andreou, Charles Lamanna, Perry Clarke and LinkedIn CEO Ryan Roslansky.
The immediate trigger is slower-than-expected Copilot adoption. Only 3.3 per cent of eligible users have signed up for the paid Copilot licence. With 6 million daily active users, Microsoft lags far behind OpenAI's ChatGPT, which counts 440 million daily users.
At a deeper level lies a strategic imperative: Microsoft wants to reduce its dependency on OpenAI. By building its own frontier models, the company gains greater control over the quality, cost and future direction of its AI platform. The OpenAI partnership remains in place, but Microsoft is diversifying its model strategy — in Wave 3, Copilot already works with both OpenAI GPT and Anthropic Claude models.
The restructuring coincides with a fundamental evolution of Copilot itself. The next generation of Copilot is designed as an agentic platform capable of executing multi-step tasks autonomously. New capabilities such as Copilot Tasks, Copilot Cowork and Agent 365 allow users and organisations to delegate complex workflows to AI agents.
Copilot Cowork, powered by Anthropic's agentic model, executes complex multi-step tasks within the Microsoft 365 ecosystem with enterprise-grade governance. Agent 365 provides a management layer that lets organisations observe, secure and enforce policy over AI agents.
For organisations already using Microsoft 365 and Copilot, there are two immediate implications. First, Copilot will continue to evolve rapidly in functionality. Second, the transition to agent technology requires a review of data governance and access management: who is authorised to let an AI agent perform actions in business systems on their behalf?
Zarioh Digital Solutions helps organisations prepare for this transition. From Copilot adoption programmes to governance frameworks for AI agents — get in touch for an exploratory conversation.
