Microsoft Puts Rival AI Models to Work Together in New Research Tool

Microsoft just made its AI assistants collaborate. The company announced a new "Critique" feature for its Copilot Researcher tool that combines OpenAI's ChatGPT and Anthropic's Claude in a single workflow. Here's how it works: GPT generates initial responses. Then Claude refines them. Microsoft says this feedback loop delivers stronger results for factual accuracy and analysis. The company also introduced "Model Council," which takes a different approach. It shows side-by-side responses from both AI models and generates a report highlighting where they agree and disagree. Users can choose either method. Both features are now available through Microsoft 365 Copilot's Frontier early access program. Microsoft claims the upgrade outperforms Perplexity's Deep Research models on accuracy and completeness benchmarks.

‍

‍