Artificial Intelligence in Excel: Copilot and AI Functions
Excel AI capabilities: what Copilot can do, how writing formulas with ChatGPT works, and what to trust it with — real scenarios and the risk boundaries.

An accountant hunts for a formula for half an hour, fails, and calls a colleague. Had they typed the same question into ChatGPT, a working answer would arrive in thirty seconds. The AI era in Excel came quietly: not with bright announcements, but with the "write this formula for me" habit.
Two separate things get mixed up in the Excel AI topic: AI inside Excel (Copilot and the smart functions), and external AI help with Excel work (getting formulas, macros and explanations from ChatGPT-type tools). The two carry different capabilities and different risks. This article separates both across real work scenarios.
What can Copilot do in Excel?
For users with a Microsoft 365 Copilot subscription, working with language inside Excel is possible: "show the 5 best-selling products in this table", "build a trend chart by month", "split the city names out of this column". Copilot reads the table, proposes a formula or a pivot, and explains the result.
The real situation is this: Copilot works well on structured, clean tables (in Table format) and gets confused in files with messy layouts. In complex multi-step analytics, verifying the result is a must. And the main barrier — the separate subscription price. For small teams, the "external AI" route below often delivers the same benefit for free.
Excel work with ChatGPT: 4 working scenarios
- Having formulas written: "Column A has dates, B has amounts. Write a formula summing the last 30 days." The answer usually works directly; when it does not, paste the error back — it will fix it.
- Having formulas explained: pasting the terrifying nested formula from someone else's file and saying "explain this in plain language". The biggest time saving with inherited files is here.
- Having macros/VBA written: "a macro that saves each sheet as a separate PDF"; the code arrives ready. Testing the code on a test file before running it is a must.
- Structure advice: "suggest a table structure for a monthly sales report"; it removes the blank-sheet fear.
One rule covers all four scenarios: describe the table's structure to the AI, not the confidential data itself. Saying "column A is the customer name, column B the debt amount" is enough; pasting real names and figures is unnecessary — and often not allowed.
Excel's built-in smart functions (without Copilot)
| Feature | What it does | Where |
|---|---|---|
| Flash Fill | Looks at your example and fills the remaining rows (like splitting first+last names) | Ctrl+E |
| Analyze Data | Ready analysis and chart suggestions over the table | Home → Analyze Data |
| Forecast Sheet | Builds a forecast from a historical series | Data → Forecast Sheet |
| Recommended Charts/PivotTables | Proposes visuals fitting the data | The Insert menu |
Flash Fill deserves a separate mention: it does the work of dozens of TEXT formulas with one example + Ctrl+E. Even without being called "AI", its daily usefulness beats many AI features.
What should you not trust AI with?
Let us draw the boundaries openly. First: the correctness of the final numbers. A formula the AI wrote can look logical yet grab the wrong range; in a critical report, hand-verify the result on a small sample. Second: confidential data. Uploading the whole finance file to an external service is a data-policy violation; work with a structure description. Third: a macro you do not understand. VBA code can delete and overwrite files; the test-copy rule is absolute here.
One observation: the users who benefit most from AI are not those who know the least, but those past the intermediate level. Because they can feel whether a result is plausible. That feeling comes from knowing the core formulas; AI does not replace it — it builds on top of it.
Frequently asked questions about Excel and AI
Is Copilot worth paying for?
For teams working hours a day in Excel and living inside the Microsoft 365 ecosystem, it is worth trying. For occasional spreadsheet users, the ChatGPT + Flash Fill combination covers most of the same need for free.
Can ChatGPT read an Excel file directly?
On the paid plans, uploading and analysing a file is possible. The privacy rule does not change: before uploading files with sensitive data, check company policy and anonymise where possible.
How do I know if the AI wrote the formula wrong?
Hand-verify the result on a small test range and try the edge cases (blank cells, zero, negative numbers). Having the formula explained is a check too: if the explanation does not square with your logic, the formula is wrong as well.
Does Google Sheets have AI?
The Gemini integration exists on the Workspace plans; the ChatGPT scenarios work identically for Sheets. The Google Sheets guide covers the platform's own strengths.
Professional support
Want to bring AI into your work processes systematically?
For diagnostics, priorities and implementation architecture, see the AI Adaptation & Strategy service.
Sources and further reading
Where to verify the source
Copilot's capabilities update quickly; check the current feature list at the official source.
- Microsoft Support: Copilot in Excel: the official capabilities and requirements
- Microsoft Support: Flash Fill: example-based filling
Continuing the topic
AI help builds on foundational knowledge; the foundations are in these articles:
- Learning Excel: a practical guide from zero
- The most important Excel formulas
- How to use ChatGPT
- The 10 golden rules of prompt writing
- Other articles on this topic
Run one experiment on tomorrow's working day: describe to the AI the formula you last struggled with. That thirty-second experience will change your relationship with Excel — in the good sense.
I'm Anar Rustamli - a strategist, entrepreneur, and AI adoption leader working at the edge of growth, technology, and human thinking. Since 2016, my work has focused on helping businesses evolve in a rapidly changing digital landscape. I design growth systems, AI-powered workflows, and strategic frameworks that align performance with purpose. I believe real growth happens when strategy, data, and human insight work together - and my mission is to help businesses adopt AI in a way that strengthens both their results and their identity.

