Microsoft Enhances Copilot with AI-Driven Research Tools

# agents

Microsoft is rolling out a new "deep research" feature in its Microsoft 365 Copilot AI chatbot, joining the likes of OpenAI's ChatGPT, Google's Gemini, and xAI's Grok in the race to enhance chatbot capabilities. These tools are powered by reasoning AI models that can tackle complex problems and even fact-check themselves, which is pretty crucial for digging deep into any topic. Microsoft's take on this includes two new tools: Researcher and Analyst. Researcher is a blend of OpenAI's deep research model, which also fuels ChatGPT's deep research tool, along with some fancy "advanced orchestration" and "deep search capabilities." Microsoft says Researcher can whip up analyses, like crafting a go-to-market strategy or putting together a quarterly report for a client. On the other hand, Analyst is built on OpenAI's o3-mini reasoning model and is all about "advanced data analysis." It works through problems step-by-step, refining its "thinking" to give you a detailed answer. Plus, Analyst can run Python to handle those tricky data queries, and it'll even show you its "work" so you can check it out. What sets Microsoft's deep research tools apart is their ability to tap into both your work data and the vast expanse of the internet. For instance, Researcher can pull in data from third-party connectors like Confluence, ServiceNow, and Salesforce, which is pretty cool. Of course, the big challenge here is making sure these tools don't start making things up or "hallucinating." Even the best models, like o3-mini, aren't perfect and can sometimes mis-cite sources, jump to wrong conclusions, or rely on questionable websites for their reasoning. To get these tools into the hands of users, Microsoft is launching a new Frontier program. Customers of Microsoft 365 Copilot who join Frontier will be the first to try out experimental Copilot features, including Researcher and Analyst, starting in April.

Related article

Comments (45)