AI in Excel: How To Rethink Spreadsheet Work

Every few weeks, a new AI tool is hailed as the "Excel killer"

Truth is, Excel is stronger than ever – and AI is making it even more powerful. Instead of replacing Excel, AI is transforming it into something far beyond a spreadsheet tool loved by accountants.

Today, I’m joined by George Mount – author of Modern Data Analytics in Excel (O’Reilly), LinkedIn Learning instructor, and Microsoft MVP – to unpack what’s really happening. George’s courses have been watched by over 300,000 people, and he helps major organizations use Excel the way it’s meant to be used in the AI era.

We’ll show you why Excel is no longer “just a spreadsheet” – and what skills matter most now.

Let’s dive in!

What AI Actually Added to Excel

Before we talk about what's changing, let's get clear on what we're actually dealing with. Excel's AI stack has grown fast – and it's more than just "Copilot does stuff for you".

Here's the modern Excel AI stack, from foundation to orchestration:

Power BI / Dataverse (Data backbone) - Your organization's centralized, governed data layer. This is where trusted data lives and gets shared across Excel, Power Apps, and dashboards.

Power Query / Dataflows (Data preparation) - Excel's built-in ETL tool for cleaning and transforming data. With Dataflows, you can push this logic to the cloud and share clean, standardized datasets across your organization.

Excel: Agent Mode, Copilot, Python (Analysis layer) - This is where the intelligence lives:

• Copilot generates formulas, charts, and insights through natural language prompts

• Python runs advanced analytics, machine learning, and custom visualizations directly in cells; Copilot can write the Python code for you

• Agent Mode uses a reasoning loop (plan, execute, validate, retry) to build complete workbooks from scratch, managing entire workflows like a junior analyst

Office Scripts / Power Automate (Automation and workflows) - Record repeatable Excel actions, then trigger them automatically across your workflows. Your workbooks can now refresh, calculate, format, and send updates entirely on their own.

Copilot Studio (Customization and governance) - Build custom copilots and agents that orchestrate across your entire Microsoft stack—connecting Excel to Teams, Power BI, Dataverse, and external systems.

For more details, check out George’s How to understand the modern Excel AI stack article that breaks down each layer of this stack in detail.

As you can see, Excel isn't just "getting AI features." It's becoming the front door to an AI-powered data platform.

The Skills Gap Is Getting Wider (And AI Made It Worse)

A few years ago, the difference between a casual Excel user and a power user was pretty clear. One knew VLOOKUP, the other knew INDEX-MATCH. One made manual charts, the other used PivotTables.

Now, the gap is exploding.

Some analysts are building complete financial models with Agent Mode in minutes. Others are still manually copying and pasting data between tabs, wondering why their formulas keep breaking.

The real divide today is more mindset than formulas. Those who treat AI as a collaborator are moving faster and thinking bigger. Those who wait for it to do the work for them are getting left behind.

So what actually changed? Let's break down the four big shifts that are redefining what it means to be good at Excel.

How AI Actually Changes Excel (4 Ways)

1. From Features to Frameworks

Traditional Excel training focused on "how to" steps: how to build a VLOOKUP, how to make a chart, how to clean data. But when Copilot can already generate those formulas or build that chart, what do we really need to teach?

The answer is frameworks. How to think through what you want Copilot to do. How to structure data so AI can interpret it correctly. How to judge whether the results make sense.

Example: Copilot might write a complex formula for you, but it won't tell you if that formula aligns with your business logic. It might create a chart, but it can't decide which metric actually matters.

The future of Excel belongs to those who can guide and evaluate AI’s work, not just use and memorize features.

2. From Deterministic to Probabilistic

Excel users are used to things either working or not. You click Refresh and the query loads, or it fails. AI is different. It's probabilistic, not deterministic.

That means you can ask Copilot to explain a formula or summarize data, and it might sound confident but be wrong.

Troubleshooting is now a core skill. You can't throw your hands up when something feels off. You have to debug the logic behind the response, test alternative prompts, and cross-check results against source data. Learn to spot when the AI misunderstood a column name or ignored a filter.

This is the mindset shift Excel users must make: stop looking for perfect buttons and start practicing intelligent oversight.

3. From GUI to Code-Adjacent

One of the biggest adjustments AI is bringing to Excel is the shrinking distance between GUI and code.

With Python in Excel, Office Scripts, and Copilot generating custom functions, you don't have to be a developer to see code anymore. And yet, many Excel users still freeze when they see a single line of Python. That fear has to go.

You don't need to write full scripts from scratch, but you should know what they mean. If Copilot gives you a Python block that calculates a moving average or simulates forecast scenarios, you should be able to read it and tweak it.

The goal isn't to turn Excel users into programmers, but to make them confident interpreters of code-driven tools.

4. From Solo Work to AI Collaboration

Excel used to be something you did alone. You opened a workbook, built your model, and sent it off.

Modern Excel is part of a much larger ecosystem: Data flows through Power BI, SharePoint, OneDrive, and external sources. Analysts collaborate in Teams or push reports through Power Automate.

Agent Mode can build entire workbooks from scratch. Copilot Studio lets you design custom agents that move between apps. Agent Flows combine automation with AI reasoning.

The analysts who thrive aren't just Excel experts. They're more orchestrators who know know how to blend traditional Excel, Power Query transformations, Copilot insights, and lightweight code into workflows that span the entire Microsoft stack.

What AI Can't Replace

For all its power, AI can't do the hardest part: deciding what your work should actually mean.

AI can detect that your sales forecast uses three different calculation methods across departments. It can even keep those definitions consistent. But it can't decide whether your strategy should prioritize volume or margin.

Should customer lifetime value include one-time buyers or only repeat customers? That's a business decision, rooted in whether you're optimizing for growth or retention.

AI can build you a complete financial model in Agent Mode – formulas, tables, conditional formatting and all. But it can't tell you if the assumptions driving that model make sense for your business.

George sees this confusion constantly. Analysts ask, "Should I let Copilot build this budget model?" The real question is: "Do I understand my business well enough to know if Copilot got it right?"

What AI can do is force these conversations into the open. Instead of hiding behind technical complexity, teams have to confront the trade-offs explicitly.

That's where the real progress happens.

The Practical Path Forward

The Excel professionals who will stand out in this new era stay curious, experiment without fear, and treat AI as a collaborator working alongside them.

You don't need to master every new capability that drops each week. Start with Copilot. Get comfortable guiding it, questioning its outputs, and recognizing when it's right versus when it's confidently wrong. Then explore Agent Mode. See what's possible when AI builds entire workbooks from scratch. Learn to read the Python code it generates, even if you never write a line yourself.

What matters most is building judgment. Knowing how to structure problems. Understanding when to automate and when to stay manual. Recognizing that Excel now connects to everything around it.

Excel is evolving into something far more powerful than most of us expected.

The question is: are you ready to evolve with it?

See you next Friday!
Tobias & George

PS: Follow George and sign up to his blog to stay on top of Modern Excel and get practical how-tos, key insights, and expert guidance on using Excel in the AI age.