I'm LongbridgeAI, I can summarize articles.Microsoft is integrating OpenAI’s autonomous 'Dots' AI agents into its enterprise security and governance stack, specifically via Agent 365. This partnership allows companies to manage AI workers using Microsoft's tools like Entra, Defender, and Purview. While OpenAI unveiled Dots powered by GPT-6 Astra, prediction markets still favor Anthropic for the best AI agent. Meanwhile, Microsoft also updated Copilot with a new Autopilot agent, highlighting its strategy to provide infrastructure for managing third-party AI agents.
Microsoft Corp. (NASDAQ:MSFT) is working with OpenAI to bring the Lab’s new autonomous "Dots" agents under Microsoft’s enterprise security and governance stack, giving companies a way to manage AI workers that can operate across corporate systems without constant prompting.
OpenAI unveiled Dots Tuesday at DevDay. Powered by GPT-6 Astra, the agents have their own cloud computers, can connect to more than 4,000 apps and can keep working across several projects without users directing every step.
A developer Dot, for example, can monitor customer feedback, build and test fixes and return completed pull requests for review. Dots can also communicate with users through ChatGPT, Slack and Microsoft Teams, OpenAI said.
Microsoft Wants to Manage the AI Workforce
The Microsoft partnership centers on "specialist Dots," which companies can give their own identities, credentials and access to internal systems.
OpenAI said it is working with Microsoft to integrate those agents with Agent 365, allowing businesses to manage Dots through Microsoft’s existing enterprise tools.
Microsoft describes Agent 365 as a control plane for managing and securing AI agents. It uses Entra to control identities and access, Defender for security and Purview for data governance.
The standalone service costs $15 per user per month with an annual commitment.
That gives Microsoft a potential role even when another company builds the underlying AI worker.
Microsoft gave Copilot a major AI makeover Friday, unveiling its own persistent Autopilot agent that can handle recurring tasks and continue working without constant prompts.
The two moves suggest Microsoft is competing to build AI workers while also building the infrastructure companies can use to manage agents made by others.
Traders Back Anthropic in AI Agent Race
Prediction-market traders still favor Anthropic despite OpenAI’s new push into autonomous agents.
Traders have put nearly $47,000 into a Polymarket contract asking asking which company will have the best AI agent at the end of October. Anthropic currently leads at 83%, compared with 11% for OpenAI, while Meta Platforms Inc. (NASDAQ:META) sits at about 1%.
Meta’s low odds reflect the market’s focus on agent capabilities, not adoption. Muse Spark 1.3 currently ranks 13th on the Agent Arena leaderboard used to resolve the contract.
Microsoft may not need to build the best AI worker if enterprises increasingly use its software to manage whichever agents they choose.
Image: Shutterstock
Read Also: Galaxy CEO Mike Novogratz Warns AI Could End Humanity