Microsoft notes progress on Copilot adoption with 15 million paid seats and around 3% penetration of commercial Microsoft 365 users, reflecting ongoing commercial rollout.

Early evaluations show the Researcher with Critique outpacing competitors on Deep Research Accuracy, Completeness, and Objectivity.

Microsoft aims to diversify away from exclusive reliance on OpenAI, deepening partnerships with multiple providers while prioritizing enterprise trust and data governance.

Work IQ powers Cowork with enterprise context and governance, ensuring secure data handling within organizations.

Microsoft positions these capabilities as enabling workers to connect steps, coordinate tasks, and follow through across workflows while staying within security and risk boundaries.

Microsoft executives emphasize the value of a multi-model approach for Copilot, highlighting orchestration and enterprise data security over any single-model capability.

The move signals a broader trend toward interoperable, multi-model AI systems that leverage strengths from different organizations rather than a single model.

Markets view enterprises want reliable, governable AI; Copilot is framed as an orchestration layer that selects the best tool for a task and adds review steps to manage risk, contrasting with rivals pushing single-model dominance.

Competition may shift toward guardrails, transparency, and governance features to manage errors and compliance in AI-assisted work.

The features are accessible in the Frontier programme as early-access tools to gather user feedback before wider rollout.

Scarlett Evans contributes reporting on these emerging technologies and related sectors.

Microsoft unveils AI upgrades to Copilot, enabling multi-model workflows via a new Critique feature that lets the Researcher agent pull outputs from OpenAI's GPT and Anthropic's Claude for each response.