Microsoft has unveiled new AI features for its Copilot assistant, designed to boost adoption and reliability. A key upgrade, called "Critique," enables Copilot's Researcher agent to utilize both OpenAI's GPT and Anthropic's Claude models simultaneously. GPT generates responses, while Claude reviews them for accuracy before presenting to the user. Microsoft plans to make this workflow bi-directional.
Nicole Herskowitz, corporate vice president of Microsoft 365 and Copilot, stated this multi-model approach will speed up workflows, curb AI hallucinations, and produce more reliable outputs. The company is also launching a 'model Council' feature for side-by-side comparison of different AI models.
These upgrades coincide with Microsoft rolling out its new Copilot Cowork agentic AI tool more widely to members in its 'Frontier' program, which provides early access to the latest features. Copilot Cowork is based on Anthropic's viral Claude Cowork product. Microsoft is racing to improve Copilot amid intense competition from rivals like Google's Gemini.