“How to use AI in Excel” is really four different questions wearing one search phrase. Most people typing it in want to know whether AI can write a formula for them, clean up a messy export, build a chart without digging through the Insert menu, or flag the one number in a sheet that doesn’t look right.
Those four things also happen to be roughly the order people discover them in. Formulas come first, because not knowing the right function name is the most obvious daily friction. Cleanup, charts, and anomaly detection tend to follow once someone realizes the same AI that wrote a formula can do more than that.
The Four Things AI Actually Does Inside a Spreadsheet
Whichever specific tool you’re using, whether that’s Microsoft’s Copilot in Excel, Google Sheets’ built-in AI, or a purpose-built AI spreadsheet, the underlying capability map looks close to the same. It breaks down into four jobs.
| What AI does | What it actually looks like |
|---|---|
| Drafts formulas from plain language | You describe the calculation you need (“show me spend divided by conversions as a percentage”) instead of hunting for a function name, and AI writes a formula draft into the cell for you to review. |
| Cleans up messy data | You point it at an import with duplicate rows, inconsistent date formats, or stray whitespace, and it standardizes the values without you writing a single cleanup formula yourself. |
| Builds charts from a description | You describe what you want compared, and AI picks a chart type and builds it, instead of you selecting a range and clicking through a chart menu by hand. |
| Spots trends and anomalies | You ask what looks unusual or what’s moving in a direction worth noticing, and AI compares rows and columns against each other instead of just repeating the numbers back to you. |
Microsoft Excel also offers AI-assisted formula workflows, including the =COPILOT() function. Behavior and plan access vary by version, so test the formula on sample cells before applying it across a workbook.
Google Sheets and other AI-enabled spreadsheet tools offer related formula and analysis workflows. Capabilities differ by product, so review generated cells and formulas before applying them broadly.
The harder part in practice usually isn’t whether any single one of these four things works. Most current tools can manage each one individually well enough. It’s whether you can chain them together on the same data, in the same place, without exporting a file, opening a separate AI chat window, and copying the results back in by hand.
Where Zoom Sheets Fits: An AI-Native Spreadsheet Inside Zoom Workplace
Zoom Sheets is one specific answer to that chaining problem. It’s a full spreadsheet, with rows, columns, formulas, pivot tables, and more than ten chart types. AI sits directly in the Zoom Sheets menu bar rather than being added as a separate add-in. Zoom Sheets is also one of four products inside the Zoom AI Productivity Suite, alongside Slides, Canvas, and Paper.
Ready to try it? Open Zoom Sheets.
Zoom Sheets can also draw on the same Zoom Workplace context where those numbers are often discussed. Meeting transcripts, decisions, and action items from a Zoom meeting or chat thread can be turned into a structured table without retyping them from scratch, which matters because many real “how do I use AI in Excel” situations start with data that came out of a call, not a blank sheet.
The rest of this article walks through what all four capabilities from the table above, formulas, cleanup, charts, and anomaly detection, actually look like back to back, inside one spreadsheet, in a single ongoing conversation with AI.
Step-by-Step: Using AI in a Real Spreadsheet in Zoom Sheets
Picture a marketing team’s campaign tracker: a handful of columns, a few months of history, and the kind of small inconsistencies every export seems to pick up along the way. Duplicate rows from a re-run report, and dates typed three different ways from one entry to the next.
That’s the exact situation this walkthrough works through.
- Start with your data. For this walkthrough, we asked Sheets to generate a sample marketing campaign dataset with columns for Campaign Name, Channel, Spend, Clicks, Conversions, and Date, and deliberately asked for a few of the small inconsistencies real exports tend to have.
Sheets came back with 20 campaign rows, plus something nobody explicitly asked for: a “Month / Spend by Month” summary and a line chart covering that same six-month span, built automatically alongside the raw data. That’s the chart-building capability from the table above showing up unprompted.
- Clean it up with one more prompt, in the same chat. Ask Sheets, in plain language, to remove the duplicate rows, standardize every date to one format, and trim any extra spaces sitting in the text fields. Sheets re-reads the sheet’s structure first, then works through the request: two duplicate rows come out, leaving 18 clean rows, and every date lands in a single YYYY-MM-DD format regardless of how it was originally typed. No separate cleanup tool, no manual find-and-replace, and no formulas of your own required.
- Add calculated columns without knowing the formula. Ask for a “Conversion Rate” column that shows conversions divided by clicks as a percentage, and a “Cost per Conversion” column that shows spend divided by conversions, specifying that you want real formulas rather than typed-in numbers.
In this walkthrough, Sheets added a Conversion Rate formula using =E2/D2 and a Cost per Conversion formula using =C2/E2, with percentage and decimal formatting applied. The formulas recalculated when the referenced inputs changed during the test. Before relying on AI-generated formulas, review the cell references, test representative rows, and confirm that each calculation matches the intended business definition.
- Ask what’s actually unusual in the data. Ask Sheets to identify any anomalies or unusual patterns, like campaigns with unusually low conversion rates or unusually high cost per conversion compared to the rest, and describe the trend in spend over the six months.
Sheets comes back with per-channel conversion rate ranges, Email between 4.6% and 5.2%, Search between 2.3% and 2.4%, Display between 1.1% and 1.5%, and Social between 0.7% and 1.4%, then flags a specific outlier by name: a campaign called “Brand Awareness Social,” with a conversion rate of 0.71%, below every other channel’s range in the sheet, and a cost per conversion around $40 that stands out from the rest of the data.
It also suggested a possible upward trend in spend and flagged what might be a data gap, noting that one month has only a single campaign logged versus several campaigns in most other months. Treat those flags as leads to investigate, not verified conclusions; compare the underlying rows and business context before acting on them.
- Ask for the exact chart you need. Rather than opening a chart menu and picking a type by hand, just describe the comparison: “add a chart comparing average conversion rate by channel.”
Sheets returns a bar chart with all four channels side by side, matching the request instead of defaulting to a generic chart type. It’s the same chart-building capability from step one, except this time it’s answering a specific question instead of illustrating the raw data.
All five of those steps happened in one spreadsheet, in one continuous chat thread, without exporting the data anywhere or opening a second tool. That continuity is what let one conversation carry the same context from raw data all the way through cleanup, formulas, analysis, and a finished chart.
After AI Generates, Cleans, or Analyzes: Editing, Sharing, and Exporting
The formulas AI writes into a Zoom Sheets spreadsheet aren’t locked or read-only. The Conversion Rate and Cost per Conversion formulas from the walkthrough above are ordinary formulas once they land in the sheet, and they can be edited, copied, or deleted by hand the same way any formula can.
Because Sheets sits inside Zoom Workplace, the same spreadsheet can be shared and edited by a teammate in real time, version history keeps a record of what changed, and permission settings control who can edit versus just view. .xlsx and .csv files shared in Zoom Chat can also be opened and edited directly inside Sheets without switching apps.
Sheets also imports and exports .xlsx and .csv files, so a spreadsheet that started as a natural-language prompt can move back into a Microsoft Excel workflow. Some formatting or formulas may vary during conversion, so open the exported file in Excel and verify it before continuing the workflow.
Bottom Line
Learning how to use AI in Excel comes down to the same four moves almost everywhere you look: writing formulas from plain language, cleaning up messy imports, building charts on request, and catching the trend or anomaly a person scanning the rows by eye would probably miss.
What changes from tool to tool is whether those four moves happen in one place, on data that’s already connected to where it came from, or whether each one means exporting, re-uploading, and re-explaining context along the way. Zoom Sheets is built for the first version: one spreadsheet, one AI menu, and a single conversation that can generate, clean, calculate, and chart your data without ever leaving the tab.
FAQ
Can AI actually write formulas for me in Excel?
Yes. Microsoft Copilot in Excel and other AI-enabled spreadsheet tools can draft formulas from plain-language requests. Review the formula, cell references, and representative outputs before applying it across a workbook.
Is there a free way to use AI in Excel?
Some tools may offer limited AI spreadsheet access, while general-purpose AI chatbots may help draft a formula or summarize a small dataset if you’re willing to copy the result back in by hand.
Zoom Sheets puts AI inside the spreadsheet toolbar, keeping formulas, cleanup, charts, and analysis connected to the working data. AI access varies by Zoom Workplace plan.
Can AI clean up messy spreadsheet data automatically?
Yes, with a prompt rather than fully on its own. In the walkthrough above, a single plain-language request covered all three cleanup tasks, removing duplicate rows, standardizing dates, and trimming extra spaces, without a separate cleanup formula or macro.
Do I need Microsoft Copilot to use AI in Excel?
No. Microsoft Copilot is one option for AI-assisted spreadsheet work, while other spreadsheet products offer their own approaches to formula generation, cleanup, visualization, and analysis. Capabilities and plan access vary by product, and AI-generated results should be reviewed before they are applied to production data.
Is Zoom Sheets the same thing as Microsoft Excel?
No. Zoom Sheets is its own spreadsheet application, one of four products inside the Zoom AI Productivity Suite, rather than a plug-in for Excel. It imports and exports .xlsx and .csv files, so data can move between the two, but the AI features described in this article are part of Sheets itself, not an Excel add-on.
Can AI in a spreadsheet spot anomalies without me asking a specific question?
Not entirely on its own. In the walkthrough above, Sheets needed a general prompt asking it to look for anomalies and trends before it produced that analysis, rather than flagging the outlier campaign automatically the moment the data changed.
Once asked, though, the prompt didn’t need to name the answer. A request pointing at the kind of thing to look for, like campaigns with unusually low conversion rates or unusually high cost per conversion, was enough for Sheets to separate a genuine outlier from an ordinary trend and explain why each one mattered.