I've taken a break from getting the Tidy Tuesday newsletter out every week, which I assume at least few of you noticed. Earlier in the year, I mentioned I might do exactly that. My activities this month have been anything but boring and useless, but I felt I didn't have something relevant enough to write without sounding forced.
This week, as February comes to an end, I still wanted to check in to report my progress with small wins (which you will, hopefully, also recall is my general focus for 2026). I also want to raise a relevant question to you all, which I'll get to in a moment.
First: I'm still working through my backlog of read-later articles. My average is below one per day, for sure—probably closer to one every three days, to be honest. But the progress is still happening, and the queue is shrinking faster than I'm adding new articles. That's a win.
I have a few other big projects I could rotate in—things I've done well with in the last year or two but haven't finished. Photos (which I hit in 2024) and my giant data archive (a main focus in 2025), for example. These projects ultimately require some big, time-consuming tasks, and that makes them psychologically difficult to go back to. But at the same time, I can do lots of little things that will ultimately move me in the same direction, and that's what I should focus on as 2026 really gets rolling.
Small wins! They might not be glamorous, but they're just as good—better, in some ways, because they're little enough to attempt in the first place.
Now, for my question. Or an intro to it, anyway.
I have been having an absolute blast recently using some AI tools—primarily Claude Code—to unleash some creativity I didn't realize I had pent up for so long. It's like having a copy of me that is way faster and better at some of the tasks I know how to do but can't devote the time to. (And, of course, plenty of tasks that I have very little knowledge about in the first place.)
I feel like I'm using these tools in a pretty balanced way; I'm not just vibe-coding my way into massive tech debt. I generally know how to do what I want, what it should look like, how the internals work, where security risks might pop up, etc. It would just take me a hundred times as long to do it by hand—and that's not an exaggeration.
For example, I threw together a little management tool for the various audio playback dashboards that allow my kids listen to audiobooks in their rooms, plus optional environmental soundscapes or white noise. I've thought about doing this for about three years. I did 95% of it last week in one evening, then polished the rest of it in short bursts over the next few nights. It does exactly what I need, and it does it well.
Another example: all that banking stuff I talked about in the last few newsletters I sent prompted me (no AI pun intended) to put together a simple "financial data flow" pipeline tool that acts as a data-cleaning middleman between banking institutions like Chase and just about anything else I might want to connect it with. It pulls the data from SimpleFIN into a flexible plaintext format, and I can just study and review it, or sort it, or filter it, or recategorize it, or build a Sankey graph, or anything—even import it into other tools. SimpleFIN is doing the really hard work of talking to the bank backend, which I can't do. But my tool gives me a platform-agnostic staging area for that data, without worrying about request limits or format changes or anything else.
Yet another example, a bit more ambitious this time: a local karate school my family attends could benefit greatly from a curriculum reference and reminder tool. So, I'm building one. Will anybody besides our family use it? Maybe, or maybe not. But the effort required to make something that works even just for us is so shockingly small that I don't care either way.
Incidentally, this is the same development process that I used toward the end of last year to create the "Tidy Organizer" local AI file analysis prototype tool I wrote about a few times.
I'm still very much aware of privacy concerns, and I'm not throwing all my personal data (including and especially financial data) into big public platforms like ChatGPT or Claude. However, I'm more than happy to use those extremely powerful platforms to build the tools that work with my data.
I would not consider myself to be an "AI expert" by any means. I'm a programmer with healthy curiosity and enough time to experiment (occasionally). Even at my level, grasping what to do with these tools—not just how to use them, but how to even imagine applying their capabilities to my life—is challenging.
It's similar to 3D printing. Owning a 3D printer gives you access to all kinds of new capabilities, but you need to be able to conceive of useful ideas in order to get actual value. 3D printers give you tremendous creative leverage, but they don't give you creativity.
AI tools likewise are an insanely powerful force multiplier: a bicycle for our minds, as Steve Jobs famously described computers many decades ago. If you look at a problem and think, "This would be so much easier if there were ten of me," then AI might just be worth applying to your situation.
…which brings me to my question.
What's one digital task you keep doing manually that feels like it should be automated?
This is not a sales pitch; I don't have some shiny new software package that I want you to buy. I'm just curious, especially after all of the truly enjoyable tinkering I've done over the last few weeks, if there are any common challenges among my readers where AI tools—custom or off-the-shelf—could genuinely help.
For example:
- Cleaning out your inbox—there's often a lot of repetition there
- Culling photos, finding less-than-great pictures or pointless screenshots
- Reading or summarizing documents
- Turning email commitments into tasks
- Categorizing or otherwise tracking financial transactions
Everyone does different things with their technology and data, and maybe you haven't ever considered whether something you do could (or should) be more efficient. Maybe you've just accepted the reality that it's a tedious task that you can't escape. But what if you could get it done with a lot less effort than you've been putting in?
Give it some thought, and let me know if you think of something. You don't have to know or even guess how it might be done with AI tools. That's for me to dig into.
That's it for now. Happy data-taming!