Excel is still the world's most loved business tool — and also where billions of hours of expensive human time get burned every year. Here are five spreadsheet tasks that AI does faster, cheaper, and with fewer errors in 2026.
I love Excel. I've used it daily for over fifteen years, and I'd still rather model a complex P&L in a spreadsheet than in any other tool. But over the past year, I've noticed something specific in client conversations: the majority of "Excel work" in mid-sized companies isn't really spreadsheet work. It's data manipulation, text formatting, and report assembly that ended up in Excel because there wasn't a better tool — until now.
This piece was sparked by Ruben's "Stop learning Excel" article, which makes a sharper version of the same argument. I'm going to focus on five concrete tasks I see teams doing manually that AI can fully or mostly handle now.
The classic example: you exported 4,000 leads from three different sources, and the names are in twelve different formats. Some are "FIRSTNAME LASTNAME". Some are "lastname, firstname". Some have phone numbers in the email field. Traditional cleanup in Excel involves regex, conditional formulas, and a lot of swearing.
Claude or ChatGPT can take a CSV like this and return clean, normalized data in 30 seconds. Even better, you can describe the rules in plain English: "Capitalize all names properly. If the email field has a phone number, move it to the phone column. Flag any rows where the company name appears to be a personal name." The model handles the edge cases your formula would have missed.
If your monthly reporting involves pulling data from Stripe, your CRM, Google Analytics, and your bank, and then assembling it into a coherent narrative — you're doing AI's job by hand. A small AI workflow can pull the data, summarize the changes, flag the anomalies, and draft the executive summary. You're not eliminating the human review; you're eliminating the eight hours of copy-paste that comes before the review.
Hundreds of survey responses, support tickets, or NPS comments sitting in a sheet, waiting for someone to manually code them into categories. This is genuinely awful work, and it's exactly the work modern language models excel at. A single prompt can produce a clean theme analysis: "The top three complaints are pricing transparency (38%), checkout flow (22%), and email frequency (14%), with representative quotes for each." You get in 90 seconds what used to take a junior analyst a week.
Project status reports, weekly client updates, sales digest emails — anything where you're transforming structured data into prose. The human still owns the judgment about what matters; the AI just writes the version that's ready to send. The interesting unlock here is consistency: every status report follows the same structure, tone, and level of detail, because the prompt enforces it.
This is the most controversial one, and I want to be careful. AI is not better than a thoughtful spreadsheet for serious financial modeling — yet. But for exploratory scenarios ("what does this look like if churn jumps 2 points?" or "what's our runway if we hire two more people in Q3?"), conversational AI is dramatically faster than building out a model from scratch. Use it for the first draft. Then move to a real model when the question is worth modeling properly.
To balance this out: there are things AI does not replace in Excel. Pivot tables on large datasets. Audit-grade financial statements. Anything where a regulator or auditor will ask you to show your work. Anything mission-critical where a hallucination would cost more than the time saved. The right framing isn't "AI replaces Excel" — it's "AI handles the messy upstream and downstream of Excel, so the spreadsheet itself can shrink to the parts where it actually shines."
My take: The biggest productivity wins I see at client companies don't come from adopting some flashy new tool. They come from auditing the manual work that crept into spreadsheets over the years and asking, task by task, whether 2026 AI handles it better. The answer is "yes" more often than people expect.
If you want a hands-on look at what's worth migrating in your specific workflow, I run a 90-minute audit session — we go through your team's actual recurring Excel work and identify the five highest-ROI things to hand to AI. You can book a discovery call here.
