Copilot in Excel: Actionable Analytics

Key insights

• Agent Mode and Analyst Agent are new Copilot experiences in Excel that use advanced models (GPT-5.2 and Claude Opus 4.5) to analyze data, suggest actions, and run multi-step workflows in your sheet.
  
• Actionable features let Copilot edit cells, build formulas, create charts and dashboards, and apply changes after you review its plan in the chat pane, so the AI both explains and executes tasks.
  
• =COPILOT() and natural language prompts enable conversational queries and dynamic results; type plain-language instructions and review the AI’s proposed steps before applying edits.
  
• Microsoft 365 Copilot licensing and a formatted table or range are required for full Agent Mode capabilities; basic Copilot features still handle simple imports and filters without the premium plan.
  
• Power Query integration and new tools like SPILL PivotTables, IMPORTTEXT/IMPORTCSV functions, and web or OneDrive imports let Copilot combine external data with spreadsheet analytics for richer insights.
  
• Automation and forecasting speed up reporting and decision-making: Copilot reduces manual work, iterates on results with follow-up prompts, and creates long-term forecasts and visual summaries from large datasets.
  

Overview of the video

In a recent YouTube video, Szymon Bochniak (365 atWork) demos Microsoft's latest evolution of Excel’s AI assistant, framing it as a more flexible and actionable Copilot. He walks viewers through new modes called Analyst Agent and Agent Mode, explaining how they layer advanced analytics and automation directly on top of workbook data. Consequently, the presentation positions these features as a major step beyond prior Excel AI tools, and the video includes timestamps to guide viewers through the highlights.

Moreover, Bochniak emphasizes that the new approach combines multilingual natural language prompts with automated actions on sheets, enabling both analysis and direct edits. He also contrasts this setup with older Python-based advanced analytics, arguing that the integrated Copilot modes are more accessible for many users. Overall, the video aims to show practical scenarios where these capabilities can speed up forecasting, visualization, and routine data tasks.

What Agent Mode and Analyst Agent do

According to the video, Agent Mode behaves like an autonomous assistant that not only suggests insights but also applies changes such as creating charts, writing formulas, and revising data. Likewise, Analyst Agent focuses on deeper analytics, letting users run forecasts, reveal factor dependencies, and extend reasoning with public knowledge to enrich internal data. Together, these modes aim to let users ask complex, iterative questions in natural language and receive both plans and executable changes.

Bochniak notes that the new Copilot modes rely on large models like GPT 5.2 and Claude Opus 4.5, and that Microsoft treats them as native Excel features starting in February. He also highlights that the tool expects structured input—formatted tables or ranges—for best results, and that the Copilot chat pane previews plans before applying edits. Thus, users retain a step to review changes even as automation increases.

How users can put it to work

The video provides practical examples: asking Copilot to build a revenue dashboard, generate a long-term forecast, or identify outliers and then apply the fixes directly in the sheet. Bochniak demonstrates iterative prompts such as refining forecasts and requesting the AI to explain formulas it generated, which helps users learn and validate results. As a result, teams can move from manual scripting or formula hunting to conversational workflows that produce dashboards and reports faster.

Furthermore, the presenter points out that premium functionality requires a Microsoft 365 Copilot license and that basic Copilot features remain available for simpler tasks. He also covers Excel-specific integrations like the new =COPILOT() function and refreshed data import dialogs, which can simplify workflows for users who regularly pull external data. Therefore, organizations should check licensing and data formatting before rolling out these features broadly.

Tradeoffs and practical challenges

While the video praises speed and automation, it also implicitly raises tradeoffs between convenience and control. For example, greater automation can save time, but it increases dependency on cloud models and vendor updates, which may complicate reproducibility and auditing of analyses. Consequently, teams that value strict versioning and repeatable pipelines may need to maintain additional documentation or controls when using Copilot-driven changes.

Additionally, Bochniak touches on data governance concerns, since the assistant reasons over workbook contents and may incorporate public knowledge, which raises questions about data residency and privacy. He warns that AI suggestions can still be incorrect or incomplete, so human review remains essential to catch hallucinations or misapplied formulas. Thus, balancing speed with careful oversight becomes a core challenge for IT and data teams.

What this means for Excel users and organizations

In summary, the video by Szymon Bochniak presents Agent Mode and Analyst Agent as tools that can change how analysts and business users interact with spreadsheets. For many users, these modes offer a faster path to insights and actionable changes, and they lower the technical bar for creating forecasts and dashboards. However, organizations must weigh licensing costs, governance needs, and training requirements before adopting the new workflows widely.

Ultimately, Bochniak’s demo suggests that Excel is shifting toward a conversational, agent-driven experience that prioritizes both analysis and action. Therefore, businesses should pilot the features with clear review processes, test for accuracy on real datasets, and align adoption with security policies to realize benefits while managing risks. As Copilot capabilities evolve, the tradeoffs between automation, transparency, and control will remain central to successful deployment.

Keywords

Excel Copilot, Copilot in Excel, Smart Copilot for Excel, Advanced analytics in Excel, Excel AI assistant, Actionable insights in Excel, Microsoft Copilot for Excel, Excel data analysis AI