Microsoft CEO Satya Nadella says AI agents are becoming a practical part of his daily workflow, especially when it comes to financial analysis, spreadsheet modeling, and tracking corporate filings.
After Intel CEO Lip-Bu Tan recently discussed using AI agents to support research and preparation, Nadella has now offered a closer look at how Microsoft’s own AI tools help him work through complex business information. In particular, he highlighted the role of Copilot’s Cowork platform and the Excel agent, which he uses to examine SEC filings, analyze spreadsheets, and continuously monitor return on invested capital, commonly known as ROIC.
Nadella explained that the real value of AI agents is no longer just about producing a simple answer or generating a finished model. Instead, he believes the next stage of AI productivity comes from being able to interact with models, challenge them, reshape them, and reason through the data behind them.
For Nadella, that means AI must do more than create a spreadsheet or summarize a document. It should help users explore why the numbers look the way they do, test different assumptions, and build multiple scenarios from the same data.
He described this shift by saying that he does not want to simply accept a model output as final. He wants to manipulate it, question it, and reason over it. That is where Microsoft’s Excel agent comes in, offering what he called a highly capable “inner loop” for spreadsheet-based analysis.
One example he shared involved a complex spreadsheet sent by Microsoft’s data center team. Rather than manually reviewing the entire file, Nadella opened the spreadsheet, selected a specific cell, and asked the Excel agent to examine the formula used there. He then instructed the agent to create five separate sheets with different scenarios based on that formula.
This type of workflow shows how AI agents could become especially useful in enterprise environments, where leaders and analysts often need to evaluate complicated models, test business assumptions, and make faster decisions from large amounts of data.
Nadella suggested that this kind of AI-powered reasoning may even resemble causal reasoning, because the software is not just calculating outputs. It is helping users understand relationships between inputs, formulas, assumptions, and results.
He also connected these capabilities to a larger question facing the technology industry: whether the massive spending on AI infrastructure can generate meaningful business returns. Microsoft and other major tech companies are investing heavily in data centers, chips, and cloud capacity to support artificial intelligence. Nadella’s comments suggest that the company sees enterprise productivity and decision-making as key areas where AI can create measurable value.
Another major advantage, according to Nadella, is the combination of internal enterprise data with broader public information. He said he uses a system that gathers SEC filing data from cloud and related infrastructure providers, organizes it into a semantic model, and then allows an AI coding agent to analyze it.
From there, the agent turns the information into a dashboard. This gives Nadella a real-time view that combines public SEC filings with Microsoft’s internal analysis, helping him track ROIC across different business layers.
In simple terms, Nadella is using AI agents not just as chatbots, but as business intelligence tools. They help him monitor market information, evaluate investments, analyze operational performance, and turn complex financial data into actionable insights.
His comments also reflect a broader trend among technology executives. Intel CEO Lip-Bu Tan has said he uses AI agents to help with research and preparation, reducing the need to rely entirely on assistants for background work. Elon Musk has also discussed the rapid development of AI agents at his own artificial intelligence company, noting strong growth in usage even as the business continues to build experience against more established competitors.
The growing interest from top executives shows how AI agents are moving from experimental tools into everyday business workflows. For companies like Microsoft, the goal is to make these agents useful inside familiar workplace applications such as Excel, Copilot, dashboards, and data platforms.
Nadella’s strongest message is that AI’s future in the workplace will depend on more than automation. The most valuable AI agents will be the ones that help people think through data, test ideas, challenge assumptions, and make better decisions faster.
As AI tools become more deeply integrated into enterprise software, spreadsheets may no longer be static files filled with formulas. They could become interactive workspaces where users ask questions, generate scenarios, review financial outcomes, and understand the logic behind complex business models in real time.
