Copilot Cowork is on stage: Autonomous artificial intelligence from Microsoft

As Microsoft continues to expand its productive AI-based business tools, it has added a new component to the Copilot ecosystem. The company opened its previously announced Copilot Cowork solution to early access within the scope of the Frontier program. In addition, notable updates have been made to the Researcher tool in Copilot. All these developments point to artificial intelligence taking on a more active role in the Microsoft 365 platform, rather than just being a supporting tool. In addition, the company’s approach to using different artificial intelligence models together is becoming more evident.

Copilot Cowork stands out as an “agentic AI” system designed especially to manage long and multi-step tasks. The system, which only asks the user to define the goal, creates a plan in line with this goal and completes the task by advancing the process step by step. Despite this, the user can intervene and guide throughout the process. In this respect, the tool aims to go beyond classical artificial intelligence assistants and work like a teammate. On the other hand, not only one-time tasks but also regularly recurring workflows can be managed by the system.

Microsoft Copilot Researcher strengthened with new features

Researcher, Microsoft’s deep research tool offered within Copilot, has also received important updates. The first of these was a new structure called “Critique”. In this system, the GPT model developed by OpenAI generates the initial response, while Anthropic’s Claude model checks this response for accuracy and quality. Thus, the content reaching the user is made more reliable. In addition, it is planned to make this process bidirectional in the future; In other words, it is on the agenda for the contents produced by Claude to be evaluated by GPT.

According to the data shared by the company, this new approach provided a significant increase in the performance of the Researcher tool. In the industry standard test called DRACO, which measures the quality of in-depth research, the vehicle’s score increased by 13.8 percent. This shows that the use of multiple models offers concrete contributions in terms of accuracy and quality. However, as debates about the reliability of artificial intelligence outputs continue, such control mechanisms are expected to gain more importance.

Another innovation was the feature called “Model Council”. This system offers the opportunity to compare responses from different artificial intelligence models side by side. In this way, users can more clearly see the differences, similarities and unique contributions of each model. In addition, it allows more conscious choices to be made in decision-making processes.

These steps by Microsoft are considered within the scope of the third wave updates of Microsoft 365 Copilot. The company aims to transform artificial intelligence from just an experimental tool into a structure that directly undertakes business processes. Despite this, how such systems will be positioned in the business world and to what extent users will have control are among the topics that will be discussed more in the coming period.