Do ideas expire, or does building this fast just mean someone else has already done it? If so, do I still need to do my version?
Being a designer has shifted and these days I am excited to be building just as much designing. It used to start with sketchpads, Figma, the visual tools I knew and then a hand off. Or I had an idea but it would take me weeks to execute and slowly code. Now the whole process happens in language — speaking, writing, thinking out loud. I’m still discovering what I’m capable of building this way.
I’ve been using Claude for the last few months — familiarizing myself with what it can do, trying things, pausing, hitting dead ends. The dead ends are part of it. That’s how you learn the edges of what the tool can actually do.
Recently I was working through a few ideas I want to execute on — a motion art series, a birdwatching archive. Claude reflected something back to me that stopped me mid-read. I’d considered the idea before, but the way it came back — shaped around my own constraints — made it land differently. I thought: I need to save this. I need to be able to come back to it.
The problem is Claude can’t find it again the same way. I’ve tried — different prompts, different angles. So I end up copying and pasting into other tools, trying to consolidate somewhere. Lately that somewhere has just been a doc.
Then I thought — what if I could just save things directly inside Claude? That would be great! So I asked. And Claude was like: nope. Your options are GitHub, a local file, or a Chrome extension you run on your own machine. Not inside here. I kept pushing to see if there was a way around it. There wasn’t.
So I thought — wait, I can actually just build this. In a few hours.
What I thought would be a quick proof of concept reminded me of something I already knew about how I work — I’m a systems thinker, and I can’t stop at the surface. Something as simple as saving turns out to not be simple at all. How do you organize what you saved? What metadata matters? How do you tag it, search it, retrieve it in a way that actually serves you later? The information architecture of a save is its own design problem. My design brain found that extremely stimulating.
What this process reminded me is that tools can help you make more tools. And increasingly I find myself wanting to build inside whatever container I’m already in — not switch contexts, just solve it now, here, in the thing. AI is collapsing the distance between having an idea and making it real. That used to take a lot longer.
The bigger reflection is this: let go and don’t hold ideas too precious. I know this already. But building at this pace makes it visceral. The idea will expire. Someone — or something — will get there. Maybe by the end of the week Claude releases something exactly like Keeps. So in the meantime I keep playing, keep learning, and keep building what I need.
If you want to try Keeps for yourself — sabrinahall.github.io/keeps