Frank
The Part Where I Admit I Did It the Hard Way
1. I spent too long learning tools instead of learning outcomes.
For the better part of three years I got deep into the mechanics of PowerShell just because I found it interesting. Syntax. Pipeline behavior. Object manipulation. Real nerd stuff. But I wasn’t building things with it consistently, I was collecting knowledge like a hobby. The scripts that actually saved people time at work came later, and they came fast once I stopped treating the language like the destination. The outcome was always the point. The tool was just the road.
If you’re learning something and you can’t name what problem it’s going to solve in the next thirty days, you might just be procrastinating with extra steps.
2. I chased platforms when I should have chased results.
I’ve stood up more home lab services than I’ll ever use. Some of them were genuinely useful. A lot of them were interesting for about a week and then became Docker containers I had to maintain for no reason. The impulse was always the same: “This could be useful someday.” Someday is a lie your ADHD brain tells you so it can have one more shiny thing to configure.
The apps I’ve actually finished: Cookslate, HookHouse-Pro, HomeBase, all started with a specific itch I needed scratched. Not “what if I built something like X.” Just, “I am personally annoyed by this problem and I’m going to fix it.” That’s the only startup energy worth trusting.
3. I underestimated how much undiagnosed AuDHD was costing me.
This one is the hardest to write because it doesn’t fit neatly into a lessons list. I’m 55. I didn’t find out about the ADHD and autism until well into my adult life. And when I look back at the years I spent feeling like I was running a race in boots that were two sizes too small, always exhausted, always behind, always confused about why other people seemed to coast through things I had to fight for, I genuinely grieve some of that time.
I’m not going to drown in the regret of it. But if I could go back five years and tell myself one thing that wasn’t about technology at all, it would be this: the way your brain works is not a character flaw. It’s a configuration. Once you know what the configuration actually is, you can start working with it instead of constantly fighting against it.
That knowledge changes everything. The problem is nobody hands it to you.
The Part Where the AI Stuff Fits In
4. I trusted AI tools based on demos instead of workflows.
When the current wave of AI assistants started getting serious, I made the mistake of evaluating them by what they could do in a single impressive response. That’s the wrong metric entirely. What matters is what the tool does on your tenth conversation, your fiftieth, when you’re in the middle of something hard and you need it to actually follow the thread.
Claude handles long context and nuance better than anything else I’ve used for coding and writing. Gemini has surprised me in HookHouse-Pro for certain music generation tasks. But I only know that because I used them for real work over months, not because I watched a YouTube demo. The demo is always the best case. Your workflow is always the real case.
5. I conflated “more automation” with “better systems.”
This one cost me probably hundreds of hours I’ll never get back. There is a version of automation that genuinely removes friction from your life. And there is a version that just moves the friction somewhere less visible while adding maintenance overhead. I’ve built both kinds.
The test I use now: if the automation broke tomorrow, would I actually notice, or would I just quietly go back to doing it manually? If the answer is “I’d go back to doing it manually,” the automation probably wasn’t solving a real problem. It was solving the problem of feeling unproductive, which is a completely different disease.
SunoHarvester is sitting unfinished on a drive somewhere because I realized halfway through that I didn’t actually need it badly enough to finish it. That’s the right answer. I wasted time building it, but I would have wasted more time maintaining something I didn’t really need.
The most dangerous kind of project is the one that’s 60% done. That’s where it’s too much work to abandon cleanly, but not useful enough to justify finishing. Learn to cut those loose faster than I did.
