I have a part-timer to thank for one of the more useful things I’ve learned about how to work.
She was sitting at a computer in the newsroom at the Statesboro Herald, working on the annual seniors section. Every spring, the paper ran a special insert with a photo of every graduating high school senior in the county — hundreds of photos, multiple schools, big community deal, logistical nightmare. Her job that day was to get every one of those photos to the same dimensions so we could lay them out. And she was doing it the only way she knew how: open image in Photoshop, click Image, click Image Size, type in the width, hit okay, save, close. Open the next one. Repeat.
I watched her do this for about four minutes before something in my brain gave out.
Not because she was doing anything wrong. She wasn’t. She just didn’t know there was a better way, and nobody had ever shown her. The tool to solve this exact problem had been sitting in Photoshop the whole time.
The rule
I don’t know exactly when I formalized it, but the version I’ve been running with for years goes like this: if I find myself doing something the same way a third time, I stop and figure out whether I can automate it. Not every time. Not a formal process. Just a three-minute pause to ask: is there a smarter version of this?
Repetitive tasks cost time, obviously, but more than that they cost attention. Every minute spent doing something the exact same way you did it yesterday is a minute your brain isn’t doing the thing it’s actually good at. The most expensive thing in any workplace isn’t the clock: it’s the part of you that’s capable of solving problems and making decisions. Spending that on tasks a machine could handle is a waste.
The part-timer wasn’t the problem. The problem was that nobody had ever asked the question on her behalf.
The Photoshop fix (ninety seconds to set up)
Photoshop has a feature called Actions. You hit record, do the thing you want to do, hit stop, and from then on you can run that exact sequence on any image, or an entire folder of images, with one click. It takes about ninety seconds to record. It runs several hundred photos in roughly the time it would take you to manually resize three.
I showed her. We recorded the action. We ran it on that year’s senior photos in about six minutes.
She had been doing this by hand for the better part of a morning.
I tell this story not to look clever but because it illustrates something about where these gaps usually live: not in complicated workflows that require engineering expertise, but in everyday tasks that people have been doing manually long enough that the manual version feels normal. The automatable version isn’t harder. It just requires somebody to notice that the repetition is happening and stop long enough to look for the tool.
Press releases and the two-spaces problem
A few years before the seniors section situation, I was doing a version of the same thing at the same paper with Word macros and press releases.
Newspapers receive a lot of press releases. Local businesses, nonprofits, government offices, school boards — everyone sends them. Some arrive reasonably formatted. Most arrive formatted for email, which means smart quotes that break when you paste them, random font size changes, the occasional subject line that made it into the body text, and — I will die on this hill — two spaces after every period. Don’t email me about it. Before you could even think about editing the content, you had to clean up the container.
So I built a macro. In Word, a macro is a recorded sequence of find-and-replace operations and formatting commands. Mine stripped the pasted formatting, standardized the quote marks, killed the double spaces, and set everything to the same font and size. Keyboard shortcut. One keystroke. Copy-paste the press release. Hit the shortcut. Actual editing could begin.
It probably saved me twenty minutes a day. Over a year, that’s roughly a week and a half of my life I didn’t spend manually chasing rogue font changes through somebody’s Chamber of Commerce announcement.
What happens when the tool builds itself
All of that was before AI made this kind of thing accessible to people without a software development background. Which is where things get genuinely interesting.
I host trivia every other Wednesday at a local pub. Have for a while. The format is consistent: I write the questions in Google Docs during the week, run the night, then post a recap video to YouTube. The YouTube side had a workflow: create the video, write a title, write a description, create a thumbnail, upload it, fill in the metadata. Every two weeks, the same steps in the same order.
This was, obviously, a “never do it twice” situation that I had been ignoring for longer than I’d like to admit.
I used Google Antigravity (Google’s agentic AI IDE, which I’ve become evangelical about) to build an app that handles the whole back end. The way Antigravity works is you describe what you want, the AI agent builds it, you test and refine, it iterates. I’m not a software engineer. I don’t need to be. What I can do is describe an output clearly: “take my question list from Google Docs, generate the video, write the title and description, create the thumbnail, and post it to YouTube.” That’s enough. The agent figures out the plumbing.
Now, when trivia night is done, I drop the question list into the app. Title, description, thumbnail, YouTube upload: done. What used to be forty-five minutes of mostly mechanical work after an already long Wednesday night takes about four.
The underlying task didn’t change. The questions still get written. Trivia still gets run. The video still goes up. What changed is which parts of that process require my actual attention and which parts just happen.
How to actually start
The first step is noticing. Repetitive tasks are invisible precisely because they’re repetitive — you do them, they’re done, you move on. They never feel like problems because they never fail. They just quietly cost time, every time, without ever raising their hand.
Pick one thing you did this week that you’ve done the same way at least twice before. It doesn’t have to be large. Reformatting a template, resizing images for social media, filling in the same fields in the same form, sending the same type of email with small variations every time. Just one thing.
Then spend ten minutes asking whether there’s a recorded action, a macro, a template, or an app that could do it for you. If you don’t know where to start, describe the problem to Antigravity, or Gemini, or whatever AI tool you’re using, and ask what’s possible. You may be surprised. The answer is sometimes “not easily” and sometimes “here’s a script that does it,” and you won’t know until you ask.
You don’t have to automate everything. You don’t have to become a person who builds apps. You just have to stop doing the same thing the same way for the third time without at least pausing to wonder if you have to.
The seniors section ran every spring. The macro ran every time a press release came in. The trivia video goes up every other Thursday morning without me thinking about it.
Nobody who reads the paper or watches the video knows how any of it got there. That’s exactly the point. The invisible infrastructure is what makes the visible work possible.
