Microsoft has introduced two groundbreaking multi-model features to its 365 Copilot Researcher tool that fundamentally change how enterprise teams approach AI-powered research. The new Critique and Council features allow researchers to run multiple AI models simultaneously, compare their outputs, and verify research quality in ways that single-model systems cannot. Critique, which uses a two-model draft-and-review pipeline, outperformed Perplexity's Claude Opus 4.6 by 13.88% on the DRACO research benchmark, marking the first time Microsoft has published a direct comparative benchmark against a named competitor in this context .

What Problem Are Critique and Council Actually Solving?

Current AI assistants operate on a single model at a time, which means they generate one response and users must trust that response is accurate. The problem is real: single-model research consistently produces confident-sounding claims that are either unsupported or drawn from low-quality sources. Critique addresses this by introducing a verification layer. The first model handles the research itself, planning the approach, sourcing relevant material, and synthesizing a draft. The second model then reviews that draft, specifically checking for source reliability, completeness, and whether claims are actually grounded in evidence rather than inference .

Council takes a different architectural approach. Rather than models working in sequence, Council runs both Anthropic and OpenAI models on the same research prompt simultaneously. A third model then acts as a judge, reviewing both outputs and generating a summary that flags where the models agree, where they diverge, and what each adds uniquely .

How to Leverage Multi-Model AI for Better Research Outcomes

• Use Critique for Verification: Deploy the two-model draft-and-review system when accuracy and source reliability are critical, such as legal research, financial analysis, or compliance documentation where errors carry significant consequences.
• Apply Council for Comparative Analysis: Run Claude and GPT simultaneously on complex research briefs to see where leading models produce different conclusions, then make informed decisions about which output better serves your specific use case.
• Implement Judge-Model Summaries: Rely on the third-party judge model to synthesize where models agree and diverge, reducing the cognitive load on researchers who would otherwise need to manually compare outputs.

Both features are currently available through the Microsoft 365 Copilot Frontier program, which targets enterprise users who need structured, sourced research outputs rather than conversational responses. Broader rollout timing has not been confirmed .

Why Is Microsoft Positioning Claude and GPT as Complementary Rather Than Competitive?

The decision to build Council into Copilot represents a meaningful product strategy shift beyond just technical architecture. Microsoft is now actively positioning Claude and GPT as complementary tools within the same enterprise workflow rather than competing alternatives. This signals that Microsoft believes no single model is definitively better across all research tasks, and that showing users the difference is more valuable than hiding it behind a single interface .

The practical implication for users is significant: they can, for the first time, see where two leading AI models produce different conclusions on the same research brief and make an informed call on which to use. This transparency approach contrasts sharply with how most AI products operate, where a single model's output is presented as the definitive answer.

However, there is a tradeoff. Enterprise users who adopted Copilot for simplicity may find a side-by-side model comparison output harder to act on than a single clean answer. Whether Frontier program participants find Council's judge-model summaries genuinely useful, or even just interesting, will determine how far this feature travels beyond the early access cohort .

Critique and Council are not incremental feature updates. They represent a fundamental change in how Microsoft thinks Copilot should operate, less as a single AI assistant and more as a structured research process with built-in verification. The question the Frontier program will answer is whether enterprise users want their AI to show its working, or whether they simply want the answer.