Microsoft announced what it calls its biggest Copilot update on Friday, with the company embedding agent runtime directly into its enterprise infrastructure and equipping every autonomous agent with persistent identity, state management, and organizational context inside Microsoft 365. CEO Satya Nadella framed the update as building Copilot into "a new OS for work" that operates across every model, form factor, and task. The release centers on Autopilot, a proactive agent designed for the enterprise that runs continuously in the background, alongside Code, which creates apps and workflows from natural language prompts.
The update introduces several structural changes for developers working with production agents. Autopilot agents receive full Entra Agent ID user accounts with productivity licenses, giving each agent its own email, calendar, OneDrive storage, Teams access, and position in the organizational chart. The agent user account sits on top of the agent identity that every Foundry agent already carries, allowing the agent to operate as itself rather than through shared service accounts or borrowed user credentials. Code generates applications that run on Microsoft Copilot Managed Runtime, now in public preview, which hosts code inside the customer's Microsoft 365 tenant boundary under IT governance. Home and Code are rolling out first through Microsoft's Frontier early-access program, while Autopilot expands to private preview at month's end.
According to the announcement, each Autopilot accepts a role and goal from the person who configures it, then keeps working in the background without needing a new prompt for each step. The report notes that a developer creates an Autopilot blueprint from a Foundry-hosted agent, which appears in the Agent 365 registry once an administrator approves it, and employees can then hire instances of that agent in Teams. The blueprint defines what the agent is designed to do, but administrators retain control over the resources and data each instance can access, applying the same access policies used for employees to agents working on their behalf. Lan Roche, Microsoft's head of global partnerships, said apps built with partner tool Lovable can now run inside a Microsoft tenant "the same way everything else does," using the same sign-in, policies, and app inventory.
The shift brings operational trade-offs. Because Microsoft controls the underlying enterprise environment, it can handle much of the work around agent state, credentials, and access controls—capabilities that independent vendors like Diagrid have been building as dedicated layers for LangGraph and other agent frameworks. But the more infrastructure a team hands over to Microsoft, the harder the agent may be to move elsewhere. The models themselves aren't locked in, since Microsoft currently runs Copilot on models from both OpenAI and Anthropic and says more labs and open-weight models are coming, and the Agent 365 SDK adds governed Model Context Protocol access to Microsoft 365 workloads regardless of framework. Long-running agents also change Copilot's economics: the standard subscription covers the assistant, but Cowork, Code, Autopilot, and other agentic features are billed based on usage through Copilot Credits, and Microsoft is extending cost management in Agent 365 to cover Code and Copilot Managed Runtime. Once an agent can keep working for hours or days without anyone watching, cost becomes part of the governance problem, which is why Microsoft is bringing those controls into the same administrative framework. For vendors building on Microsoft's stack, the convenience of offloading identity and runtime management comes at the cost of deeper architectural dependence on the Microsoft 365 ecosystem. Teams weighing that decision will need to balance near-term development speed against long-term flexibility as agent workloads scale.