|
Back from Greece, where family time (and way too much tzatziki + souvlaki) did the trick to recharge. Now that I’m back, I can see Excel didn’t rest. Some big updates rolled out while I was gone. Ones that actually make your work easier. Let’s jump into Geeky News 🤓 ✈️ AI in the grid: meet =COPILOT()Copilot in the side pane is fine. But now you can plug it directly in the grid. The new COPILOT function lets you type a natural-language prompt right into a cell and reference ranges as context. It returns AI-generated results that update automatically when your data changes. And it plays nicely with other functions, too. You can drop your results into formulas like IF, WRAPROWS, or GROUPBY for even more flexibility. Think about what this unlocks:
It’s basically like having ChatGPT wired directly into Excel’s calculation engine. Of course, there're also some "gotchas"... Math? Don't bother Large language models that power AI tools like Copilot and ChatGPT are amazing with text. But they suck at math. For numerical calculations, stick with SUM, AVERAGE, and friends. Usage limits apply
But you can get around it by passing ranges instead of running the formula row by row. Knowledge cutoff The model has no access to live web data. June 2024 is as far as it knows. It also doesn't have access to any of your organization's data, other than the context you provide directly in the formula. It's early days, and the function will only keep improving from now on. AI can make mistakes The usual AI disclaimer applies: always double-check results. Availability It's currently rolling out to Excel Insiders in the Beta channel. But you do need a Copilot for Microsoft 365 license (for business). This is the first time an Excel function requires an additional license - which honestly feels a bit strange to me. 🐍 Python in Excel can analyze your picturesPython in Excel can now directly access images in cells. You can use the Pillow library - a popular Python tool for working with images. You can inspect pixels, apply filters, adjust colors, resize, or extract metadata like resolution, format, and timestamps. All without leaving Excel. For teams handling product visuals, photo audits, or document reviews, it could add some automation muscle. For example, you can:
Now rolling out to Excel Insiders on Windows, Mac, and the Web (if your version already supports Python in Excel). 👏 Python in Excel for a competitive edgeImage processing might be a niche use of Excel, but Python in Excel really is a game changer for data analysis. Katie, who successfully completed Python in Excel For the Real World, can confirm: Huge congrats to Katie! Stories like hers are proof that while AI can speed things up, real impact comes when you understand the analysis yourself. AI in your formulas. Python for your images. Excel is stretching in directions most of us didn’t think possible. See you next week, Leila Want more?▶️ Subscribe on YouTube 🖇️ Follow us on LinkedIn 🥇 Join 400,000+ students in our courses 📣 Want to sponsor Between the Sheets? Get in touch here. 📨 If you were forwarded this message, you can get the free weekly email here. This newsletter contains affiliate links, which give us a small commission on any purchase made at no cost to you. This helps us run Between the Sheets and bring you updates like this. Thank you for your support! |
XelPlus is a leading online education company, providing training courses for Excel, Power BI, Finance, and Google Sheets. XelPlus’ bestselling courses are popular among financial analysts, CFO’s, and business owners. Technology is changing fast. We help our members turn confusion into confidence with every skill learnt.
The good thing about Excel is formulas update automatically. Change an input, the output adjusts. Ok. You know that. But what if you want the output to be a specific number. What should the input be? Let's say, your boss is standing right by your shoulder, staring at your spreadsheet. "We need to reach $2,500 in revenue this month. How many units do we need to sell?" You could keep tweaking the units cell until you get close. Guess, check, guess again. Pleasant enough if you're alone. Not so...
Most people open ChatGPT, drop in an Excel file, and type two words. Analyze this. What comes back looks good. The numbers, the insights, the tidy summary. But how do you actually know if it's right? Last week I put ChatGPT through a proper test. A messy, multi-sheet sales file with codes, lookups, and calculations it had to figure out on its own. Then I verified the results myself with a data model in Excel. There's a specific workflow that makes the difference between trusting AI blindly...
I always build my formulas the same way. I include extra rows below my data, just in case someone adds more next month. And every time, I get a column with zeros. A blank row that should be invisible suddenly shows up as zero in the spilled list. Annoying, but there are ways around it. My go-to was to wrap in a FILTER. Recently it got a lot simpler. One function. Or even better, one character. Add a dot before or after the colon in your range, like A2:.A100, and Excel quietly drops the blank...