Your Monday report wrote itself this weekend

You just opened your inbox, and it was done.

I’ve 3,131 finance pros in my community. And one of them did exactly this. He saves 10+ hours a week, because his reporting now builds itself.

You shouldn’t be doing any reactive work in 2026. So to help give you an unfair advantage, I’m running this masterclass with Spendesk. You will see how to:

• Turn plain questions into audit-ready reports in seconds (without re-formatting)
• Schedule reports to your inbox (budget vs actual, anomalies, cost-centre breakdowns, all automatic)
• Prompt with precision to get better outputs, lower your AI costs (with clean audit trials)

So to go from reactive, to having all the answers before anyone even asks. Join me here for this free masterclass before it’s full.

I was on stage in Madrid when the room went silent

I was giving an AI in Finance presentation to a group of finance pros.

I opened one Excel file. I asked Copilot (that's Microsoft's AI, built into Excel) to build a financial model from it. Then I asked it to turn that file into a clean PowerPoint, in their company colours. It took a few minutes.

The room went silent. Then someone said: "Oh, I didn't know I could do all of this."

And this is the problem. They all use Copilot, or another AI tool, every day. But they use it for one tiny thing. They don't know the rest is even there. And it's not just in Madrid. I see it everywhere. Almost no one is using AI in Excel. It’s hard to believe.

HDFC, ICICI and Axis Bank are rolling AI into their Excel and reporting teams this year.

And at the top banks, a first-year analyst costs more than $170,000 a year. Most of that year they are formatting files. They are building the same model again, and matching numbers manually. But this is the exact work AI in Excel now does for you.

So today I'll show you every main AI tool that you can use with Excel. Then I'll give you 4 levels you can work up until you've mastered it.

Your 3 Tools for Using AI in Excel

So first, your tools. And the good news: you probably already have one.

1. ​Microsoft Copilot. This is Microsoft's built-in AI. If your company runs on Microsoft 365, it is likely already there. It connects to your email, files, and company data, so it understands your work context without extra setup. Full experience requires a paid Copilot add-n starting at $21 per user/month to your existing Microsoft 365 plan.
2. ​Claude for Excel. You add this one from the Microsoft Marketplace. It sits in a side panel and is the strongest option for working across many tabs, tracing formula errors, and cleaning messy data. It requires a paid Claude plan starting at $20 per month. Your IT team may need to switch it on first.
3. ​ChatGPT for Excel. OpenAI's add-in, also from the Microsoft Marketplace. It does the job well and also works with Google Sheets, which makes it a good pick if your team splits time between both platforms. Free plans include limited access. See current pricing for paid plan options.

You also have Gemini with Google Sheets, and the ability to add ChatGPT to Google Sheets as well.

The 4-Level Ladder to Master

Level 1 - Ask for the formula. Don't Google it.

Open the AI panel in Excel. Type what you want in plain words:

It writes it for you. You don't need to remember formulas anymore.

One word of care: keep the formula simple enough that you can still check it. If it's so clever that you can't follow it, the person you send it to won't trust it either - they can't see how it works.

Level 2 - Ask it to build a template.

This is the one I use the most. Pick a report you build again and again.

My example: an aged receivables report. My accounting team used to spend hours on this. Now I ask the AI to set it up once: the age buckets, a total for everything older than 90 days, a split by supplier and by client, and the color formatting.

Most people just download a report from their system and add columns manually. A template gets the job done faster, plus it’s much higher quality in the way it looks and works as well.

Level 3 - Hand it the mess.

This is where you have super ugly workbooks with lots of messy tabs: junk rows, merged cells, different formats on every sheet etc.

Using AI in Excel, give it one clear prompt. Here’s an example from on of our AI Finance Club masterclasses:

What took an afternoon now takes about a minute.

Extra tip - To stay safe. Tell it not to touch your raw tabs. And ask it to create new tabs / sheets, plus an audit tab, so you can check the work.

For a full method on how to audit and improve accuracy with AI, you can read my two previous newsletters here and here. They’re based my 7-years audit experience at PwC.

Level 4 - Combine everything.

Here you build a full model from your own data, add the graphs, and link it all up so it stays live.

First, the rules. Don't just say "build a model." Tell it how you work first: "We use IFRS. Sales grow 30% a year. Here are my main lines." And decide up front: is this a quick one-off, or a model you'll reuse for months? If you're not clear from the start, it turns into a nightmare later.

Then say: "Build a full model from this data. Put each part on its own tab. Link them so they add up. Use formulas, not typed-in numbers. Add the graphs."

You can also link two files so the report updates itself, and build your board slides straight from the Excel file. So by next month, the numbers are already there! For a breakdown of how I create models and presentations at the same time you can read it here.

A friend of mine works at an investment firm. I thought he wasn't using AI at all. Then he showed me a full investment model he'd built with it, factoring in all of the various regulations in his industry. I was super impression. Now he tells his whole team to start every Excel job with AI.

Pro tip for Level 4

You can thank David from my AI Finance Club community for this one.

Quick tip for anyone using the Connected Files feature in the Claude Add-in for Excel.

If you ask Claude in Excel to link data between two files (for example, pulling cashflow data into a monthly reporting package), it will default to hard-coded values and tell you that linked references are not possible.

Here is the workaround:

1. Ask it to provide the cell references as text strings using an apostrophe prefix (e.g., '=Sheet1!A1).
2. Then ask it to convert those text strings into live formulas by removing the apostrophe.

That is it. Once you walk it through this sequence, it understands the capability is possible and will apply it going forward within that session.

The catch: it will forget this if you do not codify it. Turn the workaround into a skill so the behavior is repeatable across sessions, otherwise you will find yourself re-teaching it every time.

The One Thing to Remember

The room in Madrid went silent because nobody knew what they already had.

For most finance pros, the tool is right there on their screen every day (or at least a few clicks away). They just never use it.

You don't have to stay there. All you have to do it take your Excel skills, and use AI to make them even better.

That's how you go from "I didn't know I could do this" to sending your boss a report you built in five minutes - and watching them be more impressed than took all those hours doing it manually.

You’ve mastered Excel. Now it’s time to master AI in Excel ;)

Best

Your AI Finance Expert,

- Nicolas

​
​

P.S. - The best way to learn from me directly is to join my weekly masterclasses. 60-mins + Q&A where I'll get all your AI in Finance questions answered. Join me here.

P.P.S - You're on Level 1. Here's how to climb to Level 4 like a pro → How to Use Claude in Excel Like a Pro (2026)​

​