Microsoft Copilot Goes Mobile — And Wants to Do Your Work For You

Microsoft has taken its most significant step yet toward agentic AI in the enterprise. Copilot Cowork — Microsoft's AI system designed to move beyond chat and into execution, helping users delegate real tasks and have them completed — is now available on iOS and Android, allowing users to delegate work the moment they think of it, on their commute, between meetings, or away from their desk, and come back to a finished outcome.
This is not a chatbot update. This is Microsoft's vision of what AI at work actually means — and it is arriving on every smartphone.

What Copilot Cowork Actually Does

Cowork is built on Work IQ, Microsoft's intelligence layer that understands a user's data, tools and organisation, allowing it to plan, act and produce outcomes grounded in how the business runs — not just what is available on the public internet.

The distinction matters. Most AI tools answer questions based on general knowledge. Cowork operates inside a company's specific data environment — its emails, documents, calendars, CRM systems and business processes. It does not just suggest what to do. It does it.

The Numbers That Should Make Every Manager Pay Attention

Microsoft's 2026 Work Trend Index — based on analysis of more than 100,000 Copilot chats and a survey of 20,000 workers using AI across 10 countries — shows that 49% of all Copilot conversations support cognitive work: analysing, solving and thinking that once required deep expertise. More than half of AI users say they are producing work they could not have done a year ago. Among the most advanced AI users, that number reaches 80%.
Yet only 1 in 4 AI users say their leadership is clearly and consistently aligned on AI. And 65% of AI users fear falling behind if they do not adapt quickly — while almost half say it feels safer to focus on current goals than to redesign how work gets done with AI.

That gap — between what AI can do and what organisations are structured to allow — is where Microsoft is placing its biggest bet. Copilot Cowork is not just a product. It is a claim that Microsoft can bridge that gap for every enterprise customer.

Copilot Vulnerabilities — The Security Footnote

Not everything from Microsoft this week is triumphant. Microsoft disclosed and fully remediated three critical information disclosure vulnerabilities affecting Microsoft 365 Copilot and Copilot Chat in Microsoft Edge, all released on May 7, 2026. The company says no action is required from users or administrators — the patches have been applied.

But the disclosure is a reminder that as Copilot handles increasingly sensitive cognitive work, its security profile becomes increasingly critical. An AI agent with access to your emails, documents and business systems is an extraordinarily valuable target.

What This Means for GAFAM

Microsoft is executing a strategy no other GAFAM company can fully replicate: embedding AI into the productivity software that hundreds of millions of enterprise workers use every day, whether they want it or not. Google has Workspace. Apple has nothing comparable in enterprise. Meta and Amazon operate at the infrastructure layer.
Microsoft's per-seat subscription model assumes one licence for each human doing work. If AI absorbs a meaningful share of that work, that model faces structural pressure. Copilot Cowork is Microsoft's answer — charging for AI capability rather than human seats. It is a fundamental reimagining of the Microsoft business model, executed quietly, one Copilot update at a time.

The European Perspective

Organisational factors like culture, manager support and talent practices account for more than twice the reported AI impact of individual factors like mindset and behaviour. That finding has direct implications for European labour law. If AI impact depends on how organisations are structured — not just what tools are available — then the companies that redesign work around AI fastest will pull ahead. European firms, constrained by works councils, consultation requirements and regulatory caution, may find themselves structurally slower to adapt. That is not necessarily a disadvantage. It is a choice. But it is a choice with consequences. gafam.ai will be watching.