The Midnight Idea Problem

It was past midnight. I'd been coding for hours, deep in a portfolio update session, and I was tired in that specific way where you're exhausted but your brain won't stop. I grabbed my laptop and moved to bed, thinking I'd squeeze out a bit more work before calling it.

Then I needed something I didn't have. A few AI agent skills I'd set up on my Mac Mini were missing on my Macbook Air, and there was no clean way to transfer them. So I did what you do at midnight when you're too tired to think of a better solution: I reinstalled them one by one, manually, from scratch.

Somewhere in the middle of that, the thought came: what if I could just save these skills somewhere remote and pull them from any device? That was the idea that eventually became Quiver, a CLI tool for syncing and sharing agent skills across machines. I didn't open a code editor. I opened Apple Notes, wrote a few sentences, and went to sleep.

Where the good ones come from

I've started paying attention to where ideas come from, because it tells me something about whether they're worth chasing. The Quiver idea came from a genuine problem I was having in real time. I wasn't brainstorming, I wasn't trying to find a business idea, I was just annoyed. Those tend to be the ones worth writing down.

The other kind, the ideas that feel big and exciting and arrive fully formed with no friction behind them, those are fun to think about but they rarely survive contact with reality. I've chased enough of them to know the difference now, or at least to be more suspicious when an idea feels too clean.

The actual problem

Capturing ideas isn't the hard part. I have plenty of them, and Apple Notes is full of them in various states of coherence. Some are one sentence. Some have turned into pages of thinking. Some I can't even remember writing.

The hard part is knowing which ones deserve more than a note. I can only build one thing at a time, so the question I keep coming back to is which idea gets my next few months. Choosing wrong isn't a small mistake. It's months of work pointing in the wrong direction, and I've made that mistake before.

How I filter them

What I've landed on, slowly and without much ceremony, is a set of questions I ask before taking an idea any further. Did it come from something real, or just from enthusiasm? If I can't point to a specific moment where I felt the friction personally, I'm suspicious. Would I use it myself right now, not eventually, not theoretically, but today? And can I say what it does in one clear sentence? If I'm still waving my hands when I try to explain it, I don't understand it well enough yet.

Most ideas don't get through all three. Quiver did, which is part of why I built it.

Making peace with the rest

For a long time the ideas that never became anything felt like failure, like I was starting things and not finishing them. But I think the notes that go nowhere aren't waste. They're part of figuring out what actually matters. The good ideas tend to come back anyway. If Quiver hadn't surfaced that night, the frustration behind it would have shown up again eventually.

What's actually hard, and what I think about more than the ideas themselves, is the follow-through. The months of building after the initial spark, when it's no longer exciting and new but just work that needs to get done. That's the part that separates something from nothing. Not the midnight thought, but everything that comes after it.

I'm glad I didn't just close the laptop and go to sleep.