The ledger is changing again

I’ve been spending a lot of time thinking about AI this year. Not the hype cycle version. The practical version.

How does this actually fit into our business at Filterbuy? Where does it create leverage? Where is it noise?

I don’t have all the answers yet. But I’ve started to develop a framework for thinking about it, and it keeps sharpening the more I dig in.

Best practices of the past are not likely to be the best practices of the future.

That sounds obvious. But it isn’t. Because organizations are built on best practices. People get hired to execute them. Careers are built around mastering them. And when the ground shifts, the instinct is to protect what you know, not to question it. Change is hard. Letting go of the thing that made you successful is harder.

When I bought the family supply business in 2012, I inherited all the old records. Boxes of them. Recently I came across a ledger from 1989. I would have been six years old.

Every transaction. Every balance sheet entry. Every line of the income statement — written by hand. In pen. In a physical book.

In my lifetime, people went from doing their books with a pen and a ruler to QuickBooks. That’s not ancient history. That’s one generation.

People like to compare AI to the arrival of farm equipment — this massive, obvious, physical transformation. I don’t think that’s quite right. I think it’s more like the ledger becoming a spreadsheet. It doesn’t look dramatic from the outside. But the people who understood what the electronic ledger meant in 1989 built very different businesses than the people who kept buying pens.

You can already see hints of this happening with AI. A customer service rep who used to spend ten minutes searching through documentation can now get an answer in seconds. A marketing team that used to take days to draft and test ideas can now do it in hours. A developer who used to write everything from scratch now works alongside systems that help generate and debug code.

None of this looks dramatic from the outside. But inside a company, it starts to compound.

What was best practice in 1985 was not best practice in 1995. The ones who recognized that early didn’t just save time on bookkeeping. They made fundamentally better decisions because they had better tools for understanding their own numbers.

That’s where I think we are right now with AI across a lot of business functions. Customer service. Content. Internal ops. Forecasting. Code. The list keeps growing.

The real trap is waiting for certainty. Waiting until there’s a clean playbook you can copy from a competitor or a consulting deck. By the time the playbook exists, you’re already behind.

The actual work is unglamorous. It’s picking a process, understanding it deeply, and asking whether the assumptions baked into that process still hold given what these tools can now do. Sometimes the answer is no, and you rebuild. Sometimes the answer is not yet, and you move on to the next one.

We’re doing this across Filterbuy right now. Some of it is working. Some of it isn’t. I’ll share more as we learn.

But the starting point is simple: stop assuming that because something works today, it’s the right way to do it tomorrow. Look at the tools that exist now. Look at what they’ll be able to do in six months. Then plan accordingly.

The ledger is changing again.

The question is whether you’re learning the spreadsheet — or still buying pens.