Microsoft’s “Wave 3” Pivot: Agentic AI, the Anthropic Alliance, and the New E7 Suite
The landscape of enterprise software just underwent a seismic shift. Last week, Microsoft officially unveiled what it calls “Wave 3” of the Microsoft 365 Copilot rollout—a transformation that moves AI from a helpful chatbot to an autonomous colleague.
For business leaders and IT departments, the announcement marks the end of the “experimentation” phase of AI and the beginning of the “agentic” era. Here is an analysis of the three major pillars of this announcement and what they mean for the future of work.
1. Copilot Cowork: The Death of the Prompt?
The centerpiece of this update is Copilot Cowork. Developed in close collaboration with Anthropic, this feature integrates the technology behind Claude Cowork directly into the Microsoft 365 ecosystem.
Unlike standard Copilot, which requires a back-and-forth “prompt and response” cycle to create a single artifact (like an email or a slide), Cowork is designed for delegation.
- Multi-step execution: You can assign Cowork a complex project—such as “Prepare a comprehensive briefing for the Q3 regional meeting”—and it will independently pull data from Excel, summarize Teams transcripts, and draft the PowerPoint deck in the background.
- Human-in-the-loop: The system provides clear checkpoints, allowing users to review progress, steer the direction, or approve final actions before they are sent.
2. The Multi-Model Strategy: Microsoft <3 Anthropic
In a surprising strategic move, Microsoft is diversifying its “intelligence layer.” While OpenAI remains a core partner, Microsoft is now leaning heavily on Anthropic’s Claude models to power these agentic capabilities.
By adopting a multi-model architecture, Microsoft is essentially admitting that no single AI model is perfect for every task. Copilot now includes an auto-router that selects the best-suited model—whether from OpenAI, Anthropic, or Microsoft’s own Azure Foundry—depending on the complexity and nature of the request. This “model-agnostic” approach ensures that enterprise users aren’t locked into one brand of intelligence.
3. Microsoft 365 E7: The “Frontier” License
To package this new reality, Microsoft introduced its first new major enterprise license in a decade: Microsoft 365 E7 (The Frontier Suite).
Priced at $99 per user/month, E7 is the most comprehensive SKU in Microsoft history. It bundles:
- M365 E5 (Security, Compliance, and Voice)
- M365 Copilot (The AI Assistant)
- Agent 365 (The new “control plane” for managing and governing agents)
- Work IQ (The semantic graph that gives AI context on people, projects, and timelines)
The introduction of Agent 365 is particularly critical for IT leaders. As “shadow AI” (employees using unauthorized agents) becomes a risk, Agent 365 provides a central registry to observe, govern, and secure every AI agent operating within the company.
Analysis: From SaaS to “Service as a Software”
The industry is calling this shift “Service as a Software.” Historically, SaaS provided the tools for humans to do work. Now, software is beginning to perform the service itself.
The market reaction to Anthropic’s Cowork earlier this year—which saw stocks for traditional SaaS players like Salesforce and Workday take a hit—underscores the stakes. If an AI agent can independently manage a sales pipeline or coordinate HR onboarding, the value of a seat-based license for a traditional CRM or HCM tool might be called into question.
By embedding these capabilities directly into the 365 suite, Microsoft is positioning itself as the “operating system” for these agents, ensuring that even as the nature of work changes, the platform it happens on stays the same.
What’s Next?
- Late March 2026: Copilot Cowork enters Research Preview for Frontier program members.
- May 1, 2026: General Availability of the M365 E7 suite and Agent 365.
For enterprises, the message is clear: It is time to move beyond asking AI to “write an email” and start preparing for a workforce where AI agents handle the “heavy lifting” of the business process.
