I used to think spreadsheets were just something you endured. You open Excel, you brace yourself, you fight with a broken formula for twenty minutes, and eventually you get something that mostly works. Sound familiar? So when I came across this post from a founder who builds with AI, it stopped me mid-scroll.
The original poster shared a stat that honestly reframed how I look at my own workbooks: most people use less than 10% of Excel’s real capability. Less than ten percent. And according to the author, 63% of advanced users have already stopped doing Excel the manual way. They’re not typing nested formulas by hand anymore. They’re directing the software and letting AI handle the grunt work.
What I loved about this breakdown is how honest it is. This industry pro talks about building financial models by hand for hours, dealing with broken references and formula spaghetti, calling it “just Tuesday” in the early days. Now that same work takes minutes. The catch? You have to know how to prompt it. That’s the part the creator says nobody tells you early enough.
The real shift isn’t about being technical
Here’s the mindset flip the expert points to. You don’t need to memorize syntax. You don’t need to manually format anything. You describe the outcome you want, and AI executes it. The skill moves from “knowing the formula” to “knowing how to ask.” That’s a much lower barrier than most founders assume, and it’s why I think this resonates so widely.
So let me walk you through the practical playbook the post’s author laid out. I’ve turned it into a numbered list so it’s easy to come back to.
- Let Copilot edit your live workbook. The creator points out that Copilot lives right inside Excel and works on your actual sheet. It’s not a separate window you copy-paste into. It edits in place, which changes the whole rhythm of your work.
- Use it to clean, fix, and explain. According to this savvy professional, Copilot cleans messy data, repairs broken formulas, and explains errors on the spot. That last one is underrated. Instead of googling a cryptic error, you get a plain-language reason right there.
- Reach for ChatGPT on messy datasets. The author says ChatGPT is solid for generating formulas and running full analysis when your data is a mess. Great for the “I don’t even know where to start” moments.
- Use Claude for VBA, step by step. This is the tool the original poster reaches for when they need VBA explained properly. Not just the code, but the reasoning behind each step. If macros have always felt like a black box, this is your way in.
- Ask AI to identify your columns first. Before anything else, have the tool label what each column actually is. The expert stresses this keeps every downstream output accurate. Skip it, and the AI guesses, which is where errors sneak in.
- State the business outcome, not the formula. Don’t ask for a specific function. Describe what you’re trying to learn or decide. The creator’s point is that the AI often knows a better path to your goal than the formula you had in mind.
- Break big asks into steps. Instead of one giant prompt trying to do everything, this contributor recommends chopping the request into smaller pieces. You get cleaner results and it’s far easier to spot where something went sideways.
- Verify formulas against known values. Before you trust an AI-generated formula at scale, the author says test it against numbers you already know are correct. A quick sanity check now saves you from a spreadsheet built on a silent mistake.
- Never paste sensitive data into a public tool. The expert is firm on this one. Keep revenue figures and client data out of public AI tools. Protect the confidential stuff, full stop.
Why this actually matters for founders
The person who shared it says AI can cut spreadsheet work down by 40 to 60% less time. That’s not a small edge. That’s a founder getting their week back.
I think that framing is exactly right. We tend to measure AI wins in flashy demos, but the quiet victory is hours. If a task that ate half your afternoon now takes minutes, that time compounds across a month. That’s real leverage, not a party trick.
A couple of quick examples of how you might apply this yourself. Say you’ve got a raw export of sales data with inconsistent date formats and duplicate rows. Instead of manually cleaning it, you tell the AI the outcome you want (“a clean monthly revenue summary by product”) and let it propose the steps. Or maybe you inherited a workbook full of VBA nobody understands. You hand it to Claude and ask for a step-by-step explanation before you touch a single line.
The broader trend here is worth naming. AI tools are moving inside the software we already use, rather than sitting off to the side. Copilot living directly in Excel is a preview of where a lot of everyday work is heading. The founders who learn to direct these tools now are building a habit that pays off across every app they touch.
I was genuinely impressed by how clear and no-fluff this breakdown was. If you spend any real time in spreadsheets, the full post from this creator is worth a read for the details and the little nuances I couldn’t fully capture here. Check out the original on LinkedIn, and if someone on your team is still doing this the hard way, pass it along to them too.