Once upon a time, I worked in a toxicology laboratory. It was a small lab, which meant we had a small-lab budget.
We had a Laboratory Information System (LIS) that handled all the important stuff, but whenever we wanted an extra feature, the answer was usually some variation of, "Sure, we can do that. It'll cost you."
Sometimes the cost made sense. Sometimes it was cheaper to have a person do the work manually.
The task everyone hated
One of those features was document storage.
Every specimen that came into the lab had paperwork attached to it: a lab order, patient demographics, insurance information—usually two or three sheets of paper. When a specimen arrived, we'd place a barcode sticker on the paperwork that matched the specimen.
At the end of the day, all of those papers would get scanned and stored in the cloud.
The LIS company offered a feature that would automatically attach those documents to the patient's record. It sounded great. It also came with a setup fee that wasn't small and a monthly storage fee that definitely wasn't small.
So we did it ourselves.
After scanning, every file had to be renamed using a very specific format:
Patient Last Name, First Name, Date of Birth, Specimen ID
Every. Single. File.
Now, most people hated this task. I... kind of loved it. Give me a stack of paperwork, a keyboard, and an episode of Grey's Anatomy on my second monitor, and I was perfectly content.
But even I knew it was a waste of time.
So I started where every non-technical person starts
By 2018, I started wondering if there was a better way.
We already had all of the patient information inside the LIS. We could export that information to a CSV. Surely there had to be a way to connect the dots.
I had no programming background. I knew a few Excel formulas, and even then I usually had to Google them to get them exactly right.
What I did have was stubbornness.
So I started where every non-technical person starts: a spreadsheet.
After hours of digging through Microsoft forums, Reddit threads, blog posts, and countless rounds of trial and error, I eventually built a macro-enabled workbook that could take the daily export from the LIS and automatically rename every scanned file in a folder.
Instead of someone manually renaming hundreds of files, they clicked a button. That was it.
The process went from mind-numbing manual work to something that happened in seconds.
When we ran the numbers, it saved the company roughly $5,200 per year in labor costs immediately.
Not because we bought new software. Not because we hired developers. Not because we launched some revolutionary technology.
We simply looked at a repetitive task and asked:
"Why are humans still doing this?"
The desire to automate isn't new
We've always been looking for ways to save time, reduce mistakes, and spend less time on repetitive tasks.
AI just makes that easier.
That's what excites me about AI. Not because it magically does the work. Because it shortens the distance between a problem and a solution.
I gave my old problem to Claude and ChatGPT
Out of curiosity, I recently gave this exact problem to Claude and ChatGPT.
I explained the scenario and asked a simple question: How would you rename all of these files automatically?
Both of them came back with a local Python script. Which is a perfectly reasonable answer.
Except it wasn't a reasonable answer for us.
This was a laboratory full of non-technical people. Nobody was going to install Python, run scripts, or troubleshoot code when something broke.
So I pushed back. I told both tools that I needed a solution for non-technical users and that I was starting with a spreadsheet.
This time, both suggested a macro-enabled workbook.
In other words, they landed in almost the exact same place I did back in 2018.
For fun, I even had them compare my original macros to their suggestions. The solutions were remarkably similar.
The biggest difference? They added more safety checks and guardrails than I had built myself.
Where I see the value of AI today
That's exactly where I see the value of AI today.
Not because it produced a completely different answer. It didn't.
It arrived at a similar answer in minutes instead of the weeks it took me to figure it out in between my other work. It helped refine the approach. It identified potential problems. And it reduced the amount of trial-and-error required to get to a workable solution.
I'll be sharing the automations, workflows, experiments, and systems I'm building with AI—as a non-developer who likes finding better ways to get things done.
Some of them save a few minutes. Some save hours. Some save thousands of dollars. Some just do a task I hate.
But they're all built around the same question I've been asking for years:
There has to be a better way to do this, right?