Microsoft Copilot in Excel Introduces Reusable AI Skills and Live Financial Data Streams

Microsoft today rolled out a significant update to Copilot in Excel, bringing a trio of powerful new capabilities to Microsoft 365 Copilot subscribers: reusable Skills, direct financial data connectors, and workbook change tracking. The announcement, made on June 25, 2026, targets financial analysts, data professionals, and business users who rely on Excel for complex data manipulation and reporting.

At the heart of this update is the introduction of Copilot Skills, which allows users to capture, save, and replay custom AI-driven actions across any workbook. Instead of repeating the same multi-step analyses manually, users can now define a sequence of Copilot prompts and share them with colleagues. This transforms Copilot from a one-off assistant into a customizable, scalable automation engine right inside Excel.

Copilot Skills: Teaching AI to Remember and Reuse Your Workflows

The most transformative feature in today’s release is undoubtedly Copilot Skills. Think of a Skill as a macro for the AI era, but far more flexible. A financial analyst can ask Copilot to perform a series of tasks—filter a table for specific date ranges, calculate quarter-over-quarter growth, apply conditional formatting, and generate a summary chart—then save that entire sequence as a Skill named, say, “Quarterly Revenue Digest.”

Once saved, the Skill appears in a personal library and can be invoked on any comparable dataset with a single prompt: “Run Quarterly Revenue Digest on Sheet2.” Skills are reusable across workbooks and can be shared with team members, ensuring consistency in how data is processed and analyzed. Microsoft’s demo showed a Skill that automatically detects outliers in expense reports, highlights them, and drafts an email summary—all without leaving Excel.

Behind the scenes, Skills store the natural language prompts, parameter definitions (e.g., “date column”, “region filter”), and output formatting instructions. When triggered, Copilot interprets the Skill in context, dynamically adapting to the current data structure. This abstraction layer shields users from formula complexity while preserving the adaptability of AI.

Security is baked in: Skills respect user permissions and data sensitivity labels. A Skill shared by a colleague can only access data the recipient is authorized to see. Microsoft emphasized that Skills do not store actual data, only the logic—addressing early concerns about IP leakage.

Live Finance Data Connectors: Market Data Without Leaving Excel

Another headline feature is the expansion of live data connectors for finance professionals. Copilot in Excel now supports direct integration with major financial data providers, including Bloomberg, Refinitiv, S&P Capital IQ, and FactSet. Subscribers can pull real-time stock prices, historical earnings, economic indicators, and company fundamentals into any cell using natural language.

Typing “Show me the last 5 years of quarterly revenue for AAPL and MSFT, along with P/E ratios” triggers Copilot to authenticate with the configured data source, retrieve the information, and structure it neatly in a table. The data remains live: when the workbook is opened, Copilot can refresh values automatically or on demand. This eliminates the tedious process of exporting data from terminal apps and manually importing it.

For organizations concerned about compliance, the connectors respect data governance policies defined in Microsoft Purview. Administrators can control which external services are allowed, ensuring sensitive financial data doesn’t leak outside corporate boundaries.

Supported Financial Data Connectors (Initial Rollout)

Microsoft plans to expand this list throughout 2026-2027, with a public API for third-party developers to build custom connectors.

Workbook Change Tracking: Finally, a Transparent Audit Trail

Rounding out the update is a long-requested feature: workbook change tracking. Available initially on Excel for the web, this feature provides a granular, time-stamped log of every cell modification made by Copilot or by human collaborators. Users can see who changed what, when, and—if AI-driven—the original prompt that triggered the alteration.

This is a game-changer for audit-heavy environments like accounting and regulatory reporting. If Copilot suggests adjusting a revenue forecast based on new connector data, users can trace the suggestion back to the raw input and the Skill that processed it. The change log is searchable and can be exported for compliance purposes.

Crucially, change tracking differentiates between AI-suggested changes and human-accepted modifications. A green indicator shows Copilot’s proposal, while a checkmark confirms the user’s acceptance. This delineation ensures accountability remains clear.

Availability and Requirements

The new features are rolling out today to all Microsoft 365 Copilot subscribers using Excel on the web. Desktop (Windows and Mac) and mobile apps will receive the update in the coming weeks. A Copilot license is required; stand-alone Excel users without the Copilot add-on will not see these capabilities.

To access the finance data connectors, organizations must configure the desired provider via the Microsoft 365 admin center, linking corporate accounts. The free Yahoo Finance connector works out of the box for individual users.

Early Community Reaction: Excitement Tempered by Realism

Although this is a controlled announcement, early screenshots and demos have sparked lively discussion among Excel power users and financial modeling forums. Enthusiasm centers on the potential to automate repetitive monthly reporting. One portfolio analyst commented in a private beta forum: “I spend hours each week pulling data from Bloomberg into Excel. If I can just prompt Copilot and save it as a Skill, I’ll reclaim half my Fridays.”

Skepticism, however, lingers. Users recall previous AI features—like data types and Analyze Data—that fell short of expectations due to accuracy issues. “I’ll believe it when I see it handle nested IFERRORs and irregular date formats without hallucinating,” wrote a veteran modeler on a community board.

Privacy advocates raised concerns about financial data flowing through Microsoft’s AI servers. Microsoft clarified that while prompts are processed in the cloud, the connectors establish direct, encrypted tunnels to the data providers; the content of financial datasets is not used to train underlying AI models.

Hands-On: What to Expect When You First Use the Skills Panel

Microsoft provided a preview of the new Copilot pane interface. A “Skills” tab now sits alongside the chat pane, listing favorite Skills and a discoverable gallery of Microsoft-published templates. Creating a new Skill is as simple as performing a task while Copilot is active, clicking “Save as Skill,” naming it, and optionally adding a description and tags.

Advanced users can edit a Skill’s underlying prompt language—Microsoft calls it “Skill Definition Language”—to fine-tune parameters and conditional logic. This opens up opportunities for power users to build complex, parameter-driven automations without writing VBA code.

During a live demo, a Microsoft product manager showcased a Skill that ingests a CSV of customer orders, cleans missing addresses by cross-referencing a CRM data type, calculates delivery times, and flags orders at risk of delay. The entire workflow, which previously required Power Query and manual VLOOKUPs, was executed with one sentence.

Implications for Enterprise and Finance Workflows

For large organizations, these updates could reduce reliance on dedicated data integration tools. Finance departments often use tools like Alteryx or Tableau Prep to blend market data with internal spreadsheets. Copilot’s live connectors and Skills may shift some of that burden back into Excel, simplifying the tech stack.

Microsoft’s pitch is clear: AI should eliminate the “last mile” problem of getting data into the tool where decisions are made. By embedding connectors directly and making them conversational, Excel becomes a real-time analytics dashboard, not just a static spreadsheet.

Change tracking addresses a perennial audit pain point. In a post-Sarbanes-Oxley world, every material change to a financial model must be documented. Copilot’s log could satisfy auditors without manual intervention—a significant cost savings.

Getting Started and Best Practices

Microsoft recommends administrators begin by rolling out the finance connectors to a pilot group, especially those handling sensitive market data. Configuration guides are available on the Microsoft 365 admin center and via FastTrack support.

For end users, the learning curve is minimal—Skills are built from natural language, so no coding is required. Microsoft’s documentation includes video walkthroughs and sample Skill packs for common tasks like invoice processing, sales forecasting, and portfolio rebalancing.

A few tips from early adopters:
- Start small: automate a single, well-defined task before building complex Skills.
- Use descriptive names and clear descriptions for shared Skills.
- Regularly review the change tracking log to understand how AI is modifying workbooks.
- Test Skills on a copy of your data first to avoid unintentional overwrites.

What’s Next for Copilot in Excel?

Today’s release is part of Microsoft’s broader vision to make Copilot an active participant in data workflows. Roadmap hints suggest deeper integration with Power Platform—allowing Skills to call Power Automate flows—and the ability to embed Copilot chat in published Excel dashboards.

Microsoft also teased “Data Types Copilot,” an evolution of Excel’s rich data types that will let AI automatically enrich ordinary text with live attributes from authoritative sources. Combined with Skills and live connectors, this could turn Excel into a self-updating knowledge worker.

The company reiterated its commitment to security and responsible AI, stating that all features will be subject to the same compliance certifications as Microsoft 365 core services.

For now, Excel users have a compelling reason to explore Copilot. The promise of reusable, shareable AI skills and live financial data could finally bridge the gap between spreadsheet silos and real-time intelligence. The onus is now on Microsoft to deliver accuracy and reliability at scale—something that will determine whether these tools become indispensable or just another abandoned ribbon tab.