Over the weekend, my son asked if there was any work he could do to earn some money. I’ve got a few side projects running, including a Job Board for UX Researchers and Designers, and he sometimes helps to manage it: Cleaning up data, finding new jobs, getting them queued up for me to review. Just a few hours of work to buy presents, Chipotle, Tendy’s: all the things a teenager uses money for.
But instead of giving him the usual data cleanup task, I had a different idea.
Earlier that day, I’d been experimenting with an AI coding tool called Lovable, trying to re-platform the jobs board. I wanted to make some functional changes and though I’m stronger in IA/UX/IxD than development, I thought I’d use a few “vibe coding” tools to experiment and see what I could hack out.
I gave it a prompt, pointed it at the existing site, and said “build me a replacement for this.” In about four minutes, it had built a fully functioning site — complete UI, data model, everything you’d need to get the thing up and running, including one-click deploy to the cloud.
It needed a lot of work, but I was impressed by how quickly an idea became a (wobbly) working prototype.
So I told my son: “Instead of working on the Job Board, why don’t you ask Claude.ai for some instructions on how to use these AI coding tools? See if you can bootstrap your way into learning how some of them work.”
We talked about how you need a framework for testing out ideas — something you’re actually trying to build, an outcome to shoot for. It doesn’t need to be a good idea, you just need an idea.
He went upstairs and started typing away.
A few minutes later, I walked up and said, “Hey, you know how we try to plan meals every week for the three of us? No one can agree, I just grocery shop for what I think I want to make, and you guys don’t eat it and then complain.”
“Yeah,” he said.
“All right, so here’s what I want you to do: build an app that lets you invite people to suggest meal ideas. Then everybody votes on what they want for the week. Then it generates a shopping list and meal prep instructions so you can prep everything on the weekend and have nice, healthy, delicious meals that everybody’s actually going to eat.”
“Okay,” he said, and went back to work.
About an hour later he came back, grinning mischievously. “Type nomnomnomnom.lovable.app into your browser.”
I typed it in. And there was the app.
It asks about your food preferences and allergies. It lets you type in meal ideas and auto-completes them from a list of options, automatically sorted by your preferences. You invite people with a code, everybody votes on their favorites, and you get a shopping list of ingredients.
No meal-prep instructions yet, but still.
Does it need a LOT of work? Sure. Is it a fully functioning app, ready for use? Sort of — it exposes your health information to the public web if you tell it what you’re allergic to, so maybe don’t actually use it.
It’s far from a “finished working thing.” But it is a working thing. Storing data. Hosted in the cloud. With features and functions that would once have taken even a moderately skilled designer or developer days, maybe weeks, to get up and running in a prototype.
View Source → Copy/Paste App
While I may be a bit late to the game, this has opened my eyes to a new way of thinking about software development. It reminds me of how we learned to code websites in the (19)90s. If you found a website you liked and wanted to make something similar, you just clicked “View Source” in your browser. You copied the code, pasted it into a text file, and started making your own changes. You could view these changes, live, in the same browser by loading the file locally from your computer.
Fiddling with > and > tags, you bobbled your way through making changes, seeing what broke, learning the proper syntax hands-on.
You did not, in all likelihood, read a book about HTML first. You got in there, got your hands dirty, and learned by doing and making mistakes.
Down through the 25+ years since I started in the app and website development world, we’ve obfuscated so much about how things work that this mode of learning is almost impossible. Search for “how to code” or “build a website” on any search engine, and you’ll spend at least a week cutting through the ads, bogus tutorials, and AI slop.
We’ve lost the magic of Copy/Paste/Tweak.
With any of half a dozen “vibe coding” tools, you can now copy the URL of almost any public website or app and say, “build me something like that, except [blank].” It will get you maybe 80% of the way there in a couple of minutes. Then you can start refining.
Table for One
Imagine a future where software development — the nitty-gritty of ideation, planning, design, coding, testing, deployment — is as easy as view-source/copy/paste.
I can see a fairly obvious path now to a future — and really, we’re talking months, not years — where software development becomes much more like it was in the early web era: accessible, forkable, customizable.
Sure, there will still be people who develop software professionally, people who develop apps and experiences at scale. But there will also be a massive expansion in the number of people who can create software that works exactly the way they need it to work.
Not for a market segment or an organization. Not for a persona. For one person. Themselves.
On its surface, this might sound weird to anyone outside the industry. “Why would I want to build my own app?” “There are plenty of apps out there.” “I don’t have time for that. I’m not interested in that.”
But this isn’t about forcing people to do things they don’t want to do. It’s about understanding how the abstraction layer is shifting.
Right now, anyone with moderate expertise in app development can use these tools to move faster, to try out new things, to explore twenty different directions and see which one looks, or works, or feels right — all in the same amount of time they’d normally take to hammer out one idea. We can get much better ideas out the door, by leaving more of them on the cutting room floor.
As these vibe coding tools become more accessible, as they gather more feedback around how people conceive of a set of functionality we currently call “an app”, they’ll become even easier to use, until they’re so easy, they disappear.
For example: Today, I keep a list of To Dos in a Note on my iPhone. Some of those are things I do every day, so I copy them from one note to the next to always have a fresh, dated, current list of things I want to get done.
Tomorrow, Siri will “notice” that I’m doing this and say something like: “Do you want ‘copy yesterday’s to-dos’ to be part of the functionality of this app?”
I say yes. And Siri creates a new feature. She forks the Notes app that’s now personalized to me. I’m the only one using that version. And technically, I’ve “built” it.
That’s a Table for One.
-