Skip to content
Chimera readability score 0.5398 out of 100, reading level.

Microsoft 365 Copilot is your AI for work. It’s built on a multi‑model advantage—bringing the best AI innovation from across the industry into your tenant, grounded in the knowledge of Work IQ and protected with Enterprise Data Protection. Copilot is a system for work that puts that intelligence directly into the apps and workflows where work already happens, enabling you to put AI to use—from simple tasks to full, multi‑step workflows. This is a pattern that will only become more powerful as new models and new ways of working continue to emerge. Today, we are excited to announce new innovation that moves us forward on this Frontier Transformation.
We recently shared that we’re bringing the technology platform that powers Claude Cowork into Microsoft 365 Copilot. Today, Copilot Cowork—designed for long-running, multi-step work in Microsoft 365—is available via the Frontier program. Join Frontier to get early access to Microsoft’s latest AI innovations and learn more about Copilot Cowork here.
Copilot Cowork makes it easy to delegate and complete work. Describe the outcome you want, and Copilot Cowork creates a plan, reasons across your tools and files, and carries work forward with visible progress and opportunities to steer. With skills from Claude and Microsoft built in, such as calendar management and daily briefing, Copilot Cowork can handle everything from one-off tasks to repeatable workflows like a monthly budget review. Organizations like Capital Group had early access to Copilot Cowork and report already seeing its value—from planning, scheduling, and creating deliverables to preparing for executive reviews.
We have been using Copilot since its launch in 2024, and the new capabilities in Cowork will help us automate and scale the Copilot ecosystem. This isn’t about generating content or answers. It’s about taking real action—connecting steps, coordinating tasks, and following through across everyday workflows. Because Cowork operates on our enterprise data and within our security and risk boundaries, we can experiment, learn, and scale with confidence. That allows us to move faster and focus AI in places where it actually delivers value.
Barton Warner, SVP of Enterprise Technology at Capital Group.
We are also excited to announce new features in Researcher built on multi-model intelligence. Today, Researcher helps you tackle your most complex questions by synthesizing information across sources, generating comprehensive analysis, and delivering cited, well-reasoned responses you can act on with confidence.
Researcher’s new Critique feature takes this even further, using a combination of models from Frontier labs including Anthropic and OpenAI to separate generation from evaluation. One model plans the task and creates an initial draft, while a second model focuses on refinement, acting as an expert reviewer before the final report is produced.
The results are measurable—Researcher now scores 13.8% higher on the Deep Research Accuracy, Completeness, and Objectivity, or DRACO benchmark, the industry standard for deep research quality.
And with Researcher’s new model Council, you can compare responses from different models side by side, instantly seeing where they agree, where they diverge, and what each uniquely brings to the table. It’s like having multiple researchers at your fingertips. Learn more here.
Get started today
All of this innovation is part of Wave 3 of Microsoft 365 Copilot, which marks a turning point in how AI shows up at work: intelligence that understands the context of work, and trust that allows AI to scale safely across the workforce. When intelligence and trust move together, AI stops being an experiment and starts becoming how work gets done. Visit Microsoft365.com/copilot or download the Microsoft 365 app on your mobile device to get started.

Facts Only

Microsoft 365 Copilot is expanded with Copilot Cowork and new features in Researcher
Copilot Cowork designed for long-running, multi-step work in Microsoft 365
Capability to delegate and complete work by describing the outcome
Researcher's new Critique feature uses models from Frontier labs for evaluation
Improvement on Deep Research Accuracy, Completeness, and Objectivity (DRACO) benchmark
Capital Group reports benefits from using Copilot Cowork

Executive Summary

Microsoft 365 Copilot, an AI system for work, has been expanded with the addition of Copilot Cowork and new features in Researcher. Copilot Cowork allows users to delegate and complete work by describing the desired outcome, while Researcher's new Critique feature uses a combination of models from Frontier labs to separate generation from evaluation, improving its accuracy on the Deep Research Accuracy, Completeness, and Objectivity (DRACO) benchmark. Organizations like Capital Group have already seen benefits from using Copilot Cowork for planning, scheduling, creating deliverables, and preparing for executive reviews.

Full Take

This development represents a significant step in the integration of AI into daily workflows. By automating routine tasks and coordinating work across tools and files, Microsoft 365 Copilot aims to make work more efficient. The new Critique feature in Researcher demonstrates an advancement in multi-model intelligence, combining multiple AI models for better evaluation and refinement of responses. However, it's crucial to consider the ethical implications of such advanced AI systems, including data privacy, transparency, and potential misuse. As AI continues to evolve, it will be essential to maintain a balance between its benefits and potential risks.
Patterns detected: ARC-0024 Ambiguity (the article does not explicitly discuss the potential risks and ethical implications), ARC-0038 Hyperbole (improvement on DRACO benchmark by 13.8% might be perceived as significant, but context is needed to assess its actual significance)

Sentinel — Human

Confidence

This text appears to be written by a human journalist, exhibiting variable sentence length, personal voice, and unique argumentative structure. However, the low stylometric signal severity indicates a slight possibility of AI involvement.

Signals Detected
low severity: Variable sentence length and lexical diversity
medium severity: Idiosyncratic emphasis and personal voice present
low severity: Unique argumentative structure
none severity: No fabricated claims or historical inconsistencies observed
Human Indicators
The text presents a personal voice, with mentions of 'we' and quotes from Barton Warner.
The announcement of new features in Researcher is specific, citing improvements in the DRACO benchmark.