Stop waiting for the perfect tool. Build your own.
Every time something annoys me in my workflow, I don't open a tab to hunt for a SaaS anymore. I don't read comparisons. I don't sign up for a 14-day trial. I don't wait for someone's roadmap to reach where I need it.
I sit down and build.
That used to be expensive. Now it's an afternoon.
The cost of building collapsed
The math changed, and it changed in a way a lot of people still haven't caught.
What used to be a multi-week project now fits in a weekend. Sometimes it fits in a lunch break. AI doesn't write the product for you, forget that. It strips the friction out of the boring parts. Boilerplate, config, that parser you'd put off out of pure laziness. What's left is the part that matters: the idea, and the right way to solve your problem.
And here's the part that flips the switch. There's a huge band of tools that was never worth building. Too expensive to fix a problem too small. You swallowed the annoyance and moved on.
That band is gone.
The report you assemble by hand every Friday. The script you copy-paste from three different places. The spreadsheet three people update and nobody trusts. All of it just moved from "not worth it" to "I'll do it this afternoon."
I live on this. I built RTK to cut token cost across my dev operations. I built CCS to delegate tasks to other models without thinking. I built dozens of Claude Code skills that automate things I used to do by hand. None of those existed on a shelf. Because they were my problems, in my workflow, my way of working.
Do it by hand three times before automating
This is the rule that saves me from building beautiful garbage.
Don't automate up front. Run the process by hand. Once. Twice. Three times. Feel where it actually hurts, not where you assume it hurts. Only then turn it into a tool.
Sounds slow. It isn't. It's the shortcut.
Automate too early and you build the wrong solution with great care. You burn the whole weekend solving the problem you imagined, not the one you have. Then you find out the real pain was somewhere else and throw it all away.
Do it by hand first and you find the real problem before writing a single line. The three repetitions show you what repeats, what changes, where the process jams. When you finally automate, you automate the right thing.
This isn't some trick of mine. It's exactly what Anthropic recommends for building Claude Code Skills: do it by hand until the pattern is obvious, then turn it into a Skill. I follow that to the letter. Every skill I've shipped was born from a pain I'd already felt firsthand three, four, ten times. That's why they work. They didn't come from a hunch, they came from a scar.
Every tool becomes an asset
Some people look at an internal tool and see only the problem it solves. I see two more things.
First: it doesn't disappear. Third-party software changes its price, changes its terms, gets discontinued, locks you into a plan. What you build is yours. It runs your way, forever, and you own every line. It solves the problem today and keeps solving it long after the trendy SaaS hype dies.
Second, and this is the one that matters most: every tool makes you faster at building the next one. You reuse what you learned, reuse the code, reuse the instinct. The tenth tool ships in a fraction of the time the first one took. You're not just stacking software. You're stacking capability.
That's why I call it an asset, not a task. A task you finish and forget. An asset compounds.
And there's a third gain that doesn't even fit in a spreadsheet. Building is the best way to learn that exists. When you build the tool, you're forced to understand the problem down to the bone. The internals, the why behind every decision, where it breaks and why. You can't fake understanding. The code won't let you.
I've always said whoever teaches learns double. But whoever builds learns triple. Because AI won't forgive a half-understood concept, it just hands you back a bug.
Stop waiting
This is the brutal summary.
Don't wait for the perfect tool. It's not coming, and if it does, it comes expensive and shaped like someone else. Don't outsource your own workflow to a roadmap that isn't yours.
The cost dropped. The rule is simple. The asset stays. All that's left is for you to sit down and build it.
follow me and I'll show you the tools I've built and how I build the next ones.