Amid artificial intelligence (AI) turf wars and benchmark braggadocio, a new battle line has emerged: Deep- reasoning agents.
On Tuesday, Microsoft Corp. joined an ever-expanding fray of contenders with two new agents that bring multi-step reasoning to Microsoft 365 Copilot, its AI chatbot.
The new agents, Researcher and Analyst, offer higher-level research and data analysis using existing OpenAI models to deliver more advanced, personalized support. Microsoft says new agent flows in Copilot are now powerful enough to “automate any task you can imagine” with rule-based workflows that include AI actions.
Researcher relies on OpenAI’s deep research AI model for “complex, multi-step research” such as the ability to sift through mountains of information on the web and then present that information in a spiffy, sourced report. Researcher also has access to third-party data via connectors to Salesforce and ServiceNow so that business customers can glean insights across those companies’ tools. (OpenAI’s deep-research model, which powers the company’s ChatGPT deep research tool, boasts “advanced orchestration” and “deep search capabilities.”)
Additionally, Microsoft claims Researcher can perform analysis of a go-to-market strategy for example, and create comprehensive quarterly reports for clients combining information from their apps and external news sites.
Analyst, based on OpenAI’s o3-mini reasoning model and “optimized to do advanced data analysis,” is capable of chain-of-thought reasoning, according to Microsoft. The agent can turn raw data into spreadsheets, run Python to tackle complex data queries while exposing its “work” for inspection, and generally operate on the level of a skilled data scientist.
Both agents are scheduled to be available to Microsoft 365 Copilot license holders via an early access program in April, along with new autonomous agentic capabilities coming to Copilot Studio. The rollout is part of a new Frontier program that lets customers check out early Copilot innovations while they’re still under development.
Microsoft’s deep-reasoning agents follow the latest announcements from OpenAI’s ChatGPT, Alphabet Inc.’s Google Gemini, and xAI’s Grok, to bring research agents to chatbots. Those agents possess the ability to perform in-depth research by thinking through problems and fact-checking themselves.
The software giant’s foray into deep-reasoning agents could portend another potential clash with OpenAI over future AI business. It is another wrinkle in a star-crossed relationship between the two companies — from allies to fierce competitors.
Though Microsoft has sunk at least $13 billion into OpenAI, it increasingly finds itself vying for the same customers as the Sam Altman-led startup, as the AI industry explodes into a multi-trillion-dollar market worldwide.
