Microsoft is adding two new AI reasoning agents to its Microsoft 365 Copilot suite: Researcher and Analyst. These AI agents are designed to streamline workflows by handling multi-step processes that typically require significant time and expertise.
The tech giant introduced these reasoning agents, which Microsoft says are “first-of-their-kind,” as part of its continued push to make AI a core part of productivity tools.
Smarter AI for research and data analysis
Researcher
Researcher is built to handle in-depth, multi-step research projects. It combines OpenAI’s deep research model with Microsoft 365 Copilot’s advanced orchestration and deep search capabilities.
It pulls information from a mix of sources, including Microsoft 365 work data (emails, meetings, chats, and files) and external platforms like Salesforce, ServiceNow, and Confluence. Microsoft says Researcher can help users build go-to-market strategies, analyze industry trends, or create in-depth client reports by integrating internal and third-party data.
Analyst
Analyst focuses on data analysis and operates like a skilled data scientist; it processes raw data, identifies patterns, and generates insights. It’s powered by OpenAI’s o3-mini reasoning model and can perform advanced data analysis, including Python-based calculations and visualizations.
Microsoft says users will be able to watch the AI generate Python code in real time, making it easier to understand and verify results. Analyst follows a “chain-of-thought” reasoning process, meaning it works through problems iteratively to refine its conclusions, mimicking human analytical thinking.
Early access — but not for everyone
These tools will debut in April as part of Microsoft’s “Frontier” program, an early-access initiative for Copilot users. Businesses with a Microsoft 365 Copilot license can try these AI tools, while also acting as testers for features still in development.
In addition, Microsoft is rolling out updates to Copilot Studio, its tool for building custom AI agents. It’s offering new “deep reasoning” and automation capabilities that enable companies to create, manage, and deploy agents that handle multi-step tasks. The company says this will allow organizations to develop AI-driven workflows tailored to specific business needs, while also using flows to automate processes quickly and predictably.