Following announcements from Google and OpenAI regarding advancements in AI, Microsoft has made significant updates of its own. It recently introduced two pioneering “deep reasoning” agents for Microsoft 365 Copilot, named Researcher and Analyst, which are touted as the first of their kind. These new capabilities are designed to enhance the functionality of custom AI agents.
The Researcher agent utilizes OpenAI’s sophisticated deep research AI model for executing “complex, multi-step research.” It also connects to third-party data sources like Salesforce and ServiceNow, allowing business users to gain insights from various tools they employ.
On the other hand, the Analyst agent operates on OpenAI’s o3-mini reasoning model. It employs chain-of-thought reasoning to transform raw data into organized spreadsheets and execute Python code in real-time, functioning at the level of a skilled data scientist. This agent can compile comprehensive reports, similar to this example (pdf).
These groundbreaking tools are expected to begin rollout in April for Microsoft 365 Copilot license holders through an early access program. Additionally, new autonomous agent capabilities are being introduced in Copilot Studio.
Microsoft asserts that the new agent flows within Copilot are powerful enough to “automate any task you can imagine,” employing rule-based workflows that incorporate AI actions. The LinkedIn announcement illustrates scenarios where agent flows can direct feedback to the appropriate team, though we will need to assess its practical application compared to traditional methods, such as checkboxes, and evaluate the effectiveness of its “low code” experience.
