Microsoft's Fix for AI 'Double Agents' Will Cost You $99 a Month

Microsoft is rolling out its most expensive enterprise software tier yet, Microsoft 365 E7, packaging its new AI governance tools with a warning that ungoverned AI could act as corporate “double agents.” According to VentureBeat, the price for its fix, a new product called Agent 365, is $99 a month. The move signals a clear strategy: introduce more powerful, autonomous AI capabilities and then sell the expensive tools required to control them.

Both the new E7 tier and Agent 365 will be generally available on May 1st, as reported by VentureBeat. Microsoft’s marketing frames this as a necessary security measure for the modern enterprise. The company is cautioning customers about the risks of deploying autonomous AI agents without a strong governance framework, coining a memorable threat to spur adoption. The core of this strategy is to get more businesses to adopt AI by bundling it with the identity, management, and security features that large IT departments demand, as CNBC noted.

A Premium Solution for a Self-Made Problem

Microsoft's product announcements appear to be working in tandem. On one hand, the company is introducing “Copilot Cowork,” a new agentic AI tool designed to complete work on a user's behalf across multiple Microsoft 365 applications, according to a separate VentureBeat report. This tool represents a significant step up in autonomy for Copilot, moving it from a helpful assistant to an active digital colleague.

On the other hand, Microsoft is launching Agent 365 to rein in the very type of technology that Copilot Cowork represents. By creating more powerful agents, the company also creates a more urgent need for the security and governance tools included in its highest-priced E7 bundle. This pattern indicates a classic enterprise upsell strategy, where new capabilities are paired with premium management tools designed to increase average revenue per user. The goal, as CNBC reports, is to juice sales with AI.

An Unexpected Anthropic Alliance

A surprising detail in the announcement of Copilot Cowork is the involvement of Anthropic. VentureBeat reports that Microsoft is getting “help from Anthropic” for this new agentic tool. This is a notable development given Microsoft’s multi-billion dollar investment in and deep integration with OpenAI. For Microsoft to collaborate with Anthropic, one of OpenAI’s primary competitors, suggests a strategic hedge.

This analysis points to a multi-vendor approach to its AI future. By working with Anthropic, Microsoft diversifies its AI model sourcing, reducing its dependence on a single partner. It also allows the company to leverage different models for different tasks, potentially using Anthropic’s models, known for their focus on safety and reliability, for sensitive enterprise agentic tasks. This move introduces a new dynamic into the platform wars, where even tightly-allied partners are not exclusive.