Month-end is super stressful (some of you still haven’t closed and we are on day 12 already). Even if your process is simple, you’re probably still spending 8 or more hours every month doing the exact same sequence. Downloading from your accounting system → Opening Excel → Mapping accounts → Filtering → Copying to actuals. Pulling budgets → Calculating variances → Formatting. (It’s boring just writing this) Plus, this is before you even start to write your commentary for your management packs. And long before you get the chance to do any analysis that will help you make an impact. There are so many steps where your brain is not involved. And, now I am going to say something I never say. You are probably using AI too much 😉 Today I will show you where AI can become a problem, plus a method you can use today to free up your time for activities that create more of a business impact.
|
|
|
In simple terms:
- Probabilistic tools like ChatGPT or Copilot give you likely answers based on patterns in data, so the same question can produce slightly different outputs each time.
- Deterministic tools like formulas, scripts, and automations give you the same answer every time when given the same inputs.
For repeatable finance work (like BVA prep, mappings, consolidations), you want deterministic systems so your numbers are consistent, auditable, and easy to trust.
Note: BVA is often used in the US or UK as an abbreviation for Budget Versus Actuals).
Build once, run Forever
|
|
Instead of thinking about ‘more AI’ you need to be thinking about ‘more automation’.
The sort of thing that you can ask AI to help you build ONCE.
And then have it save you time forever.
When we talk about doing your BVA, your budget versus actuals, your analysis. Most of it is actually a calculation. You don’t need AI for this.
We need AI to write the calculation, or the automation once, and then the automation runs itself.
This can be as simple as a script that you paste into Office, or into Google Workspace.
And you would be surprised how much time a simple script can save.
In the case of repeat activities, think of AI as a developer you hire for one project. They build the automation, hand it over, and you never need to call them again.
Then you hire your next AI developer to build another project, and you keep going until you’ve saved enough time to block out an entire Wednesday afternoon to teach your kids French 😉 (If you did not guess, this is what I do).
So how do you do this?
How to do this (today)
Step 1 – Tell AI what you do manually
Open ChatGPT (or Copilot, or Gemini) and describe one repetitive task in detail.
Step 2 – Ask AI to write the code
Tell AI you want to automate this and ask it to write the script. You only need to do this once.
If you use Microsoft Excel, ask for VBA code or Office Script. If you use Google Sheets, ask for Google Apps Script.
AI will write the code for you. You paste it into your spreadsheet. You run it. If something doesn’t work, tell AI what went wrong and it will fix the code. Keep iterating until it works.
Step 3 – Add verification
Before you finish, ask AI to add checkpoints to the code so you’re not trusting the automation blindly.
Where to run your scripts
Office Scripts (Excel for web)
|
|
- Open your file in Excel Online
- Click Automate in the ribbon → New Script
- Paste your code in the Code Editor
- Click Run
VBA (Excel desktop)
|
|
- Enable the Developer tab: File → Options → Customize Ribbon → tick Developer
- Click Developer → Visual Basic (or press Alt+F11)
- Click Insert → Module, then paste your code
- Press F5 to run (or close the editor and use Developer → Macros)
Google Apps Script (Google Sheets)
|
|
- Open your Google Sheet
- Click Extensions → Apps Script
- Delete any placeholder code and paste yours
- Click the ▶ Run button
Pro Tip – Ask AI to not just create the script, but give you step by step instructions on how to run it as well.
Finance use cases you can automate this week
Consolidating department files
Mapping account codes
Splitting and emailing reports
Budget vs Actuals prep
The one thing to remember
AI is probabilistic. Automation is deterministic.
When you ask AI to help with your reporting every month, you’re having a conversation that will go slightly differently each time.
When you ask AI to write you an automation once, you’re building something that will work exactly the same way every time you run it.
That’s the difference between adding AI to your workflow and using AI to eliminate the workflow.
So pick one task this week. The one that makes you think “I cannot believe I’m still doing this manually.” Describe it to AI, ask for the code, paste it, run it, iterate until it works.
You might spend 2 hours getting it right. But those 2 hours buy back 30 minutes every month for the rest of your career.
And next month, when your team asks “when will the reporting be ready?” you’ll be able to say something different.
“It’s already done. Let’s talk about what the numbers mean.”
Best,
Your AI Finance Expert,
Nicolas
P.S. – What did you think of this? Will you use it? Hit reply (I read all replies)
P.P.S. – I have talked about this in a video I made last year. Showing the difference between automation, workflow, and agent. If you want to see more automation use cases watch my full video here.
|
|