Skip to main content
Microsoft Copilot with Autopilot Agent, showing usage-based billing interface on a computer screen.

Editorial illustration for Microsoft Adds Autopilot Agent to Copilot, Enables Usage-Based Billing

Microsoft Copilot Gets AI Agent Autopilot Feature

Microsoft Adds Autopilot Agent to Copilot, Enables Usage-Based Billing

• 4 min read

Microsoft rebuilt Copilot again, this time splitting the app into three distinct sections and handing businesses an AI agent that runs on its own even when nobody's watching. The centerpiece is Autopilot, a rebrand of the previously announced "Scout," built on OpenClaw. Each Autopilot instance comes with its own cloud computer, its own storage, its own identity, letting it sit inside a Teams channel and coordinate supplier evaluations or grind through recurring tasks without a human clicking anything.

Employees can summon it with an @mention in Teams, Outlook, or a document, then log off. The agent keeps working.

The other two panels, Home and Code, push Copilot further into territory Microsoft has been circling for a while. Code opens app-building and dashboard creation to people who've never written a line of software, running on the same infrastructure as GitHub Copilot. Home folds chat and collaboration together, with a "Today" dashboard set to hit private preview in October.

Underneath all of it, Microsoft is also tearing up how it charges for the product, moving away from flat-rate licensing toward a model that tracks usage instead.

Microsoft is shifting Autopilot, Code, and Cowork to usage-based billing instead of a flat rate, similar to how ChatGPT Enterprise works beyond a certain quota.

Why this matters

Microsoft is betting that IT departments want an agent living inside their Teams chats and file systems around the clock, not just a chatbot that answers when asked. The rebrand from "Scout" to "Autopilot" is telling: Microsoft wants this framed as infrastructure, something that runs quietly in the background rather than a tool you consciously invoke. For developers and founders building on Microsoft's stack, the per-instance cloud computer and identity model is worth watching closely, because it changes the cost and security calculus.

Usage-based billing on an always-on agent means expenses can climb fast if governance controls aren't tuned correctly, and Jared Spataro's emphasis on permissions and auditing suggests Microsoft knows enterprise buyers will ask hard questions before deployment. The "Code" section, letting non-developers build apps and automations in natural language, is the more disruptive piece long-term. It pushes Copilot further into territory once owned by dedicated no-code platforms.

Whether "Microsoft IQ" actually understands organizational context well enough to earn broad, unsupervised access to Teams and files, or whether this becomes another over-permissioned assistant, will decide if Autopilot is a genuine productivity shift or just Scout with a new name and a bigger invoice.

Common Questions Answered

What is Autopilot and how does it differ from the previous Copilot design?

Autopilot is Microsoft's rebrand of the previously announced "Scout" agent, built on OpenClaw technology. Unlike the traditional Copilot chatbot that responds when users ask questions, Autopilot runs autonomously within Teams channels with its own cloud computer, storage, and identity, enabling it to coordinate tasks and handle recurring workflows without human intervention.

How does the new usage-based billing model work for Autopilot, Code, and Cowork?

Microsoft has shifted Autopilot, Code, and Cowork from flat-rate pricing to usage-based billing, similar to how ChatGPT Enterprise operates beyond certain quotas. This means businesses will pay based on actual consumption rather than a fixed monthly fee, allowing for more flexible cost management depending on usage levels.

What are the key capabilities of each Autopilot instance?

Each Autopilot instance comes equipped with its own dedicated cloud computer, independent storage, and unique identity, allowing it to operate seamlessly within Teams channels. These instances can coordinate supplier evaluations, automate recurring tasks, and manage complex workflows continuously without requiring manual user interaction or oversight.

Why is Microsoft's rebrand from Scout to Autopilot significant?

The rebrand from "Scout" to "Autopilot" reflects Microsoft's strategic positioning of this technology as background infrastructure rather than a conscious tool users actively invoke. This framing emphasizes that the agent is designed to run quietly and continuously in the background, fundamentally changing how businesses think about AI assistance from reactive chatbot to proactive autonomous agent.

What does the per-instance cloud computer and identity model mean for developers?

The per-instance cloud computer and identity model provides developers and founders building on Microsoft's stack with a more sophisticated infrastructure approach to AI agents. This architecture enables each Autopilot instance to maintain its own isolated environment and identity, allowing for better security, scalability, and customization for enterprise deployments.

LIVE23:54Court Denies Anthropic's Bid to Block Trump Blacklisting