To ship its latest Copilot features, Microsoft is leaning on both OpenAI and Anthropic, betting its advantage is not in models but data.

The New Stack is a publication covering trends and technologies in cloud-native development, DevOps, and software delivery. Developers can learn about containerization, Kubernetes, and cloud computing, as well as explore topics such as microservices architecture, serverless computing, and continuous integration/continuous delivery (CI/CD) pipelines.

The New Stack

Microsoft is now using both OpenAI's GPT and Anthropic's Claude in tandem within its Copilot Researcher agent. A new 'critique' feature has GPT draft responses while Claude reviews them for accuracy, completeness, and citation integrity — a multi-model approach that outperforms either model alone on the DRACO benchmark (57.4 vs. 50.4 for Claude in Deep Research mode). Microsoft is also bringing Anthropic's Cowork tool (long-running multi-step agents for knowledge workers) into Copilot under the name Copilot Cowork, now available in the M365 Frontier early-access program. The move reflects Microsoft's strategic bet that its value lies not in owning frontier models but in the enterprise data and trust layer it provides around them, while simultaneously deepening its dependency on external model providers.

Microsoft’s Copilot makes Anthropic’s Claude and OpenAI’s GPT team up

Cowork is now in the M365 Frontier Program