Microsoft is set to launch a “deep research” AI tool within Microsoft 365 Copilot, its AI chatbot application.
Recently, numerous deep research agents have been introduced across various chatbots, including OpenAI’s ChatGPT, Google’s Gemini, and xAI’s Grok. These are powered by reasoning AI models capable of problem-solving and self-fact-checking, which are crucial abilities for conducting comprehensive research.
Microsoft’s offerings are named Researcher and Analyst. Researcher integrates OpenAI’s deep research model, which also powers ChatGPT’s deep research capabilities, with advanced orchestration and deep search functionalities. According to Microsoft, Researcher can perform tasks such as developing go-to-market strategies and creating quarterly client reports.
The tool named Analyst utilizes OpenAI’s o3-mini reasoning model, optimized for advanced data analysis. Analyst operates by iteratively addressing problems, refining its approach to provide detailed answers to inquiries. It can execute the Python programming language to handle intricate data queries and makes its analytical process open to inspection.
What sets Microsoft’s deep research tools apart from their competitors is their capability to access both organizational data and internet resources. For instance, Researcher can use third-party data connectors to leverage information from AI agents, tools, and applications like Confluence, ServiceNow, and Salesforce.
A significant challenge is ensuring that tools like Researcher and Analyst do not produce incorrect data or conclusions. Models such as o3-mini and deep research, while advanced, are not infallible and can occasionally misattribute sources, draw erroneous conclusions, or rely on unreliable public websites.
Microsoft has initiated a new Frontier program through which Microsoft 365 Copilot users can access Researcher and Analyst. Participants in the Frontier program, which will be the first to receive experimental Copilot features, are expected to have access to these tools starting in April.