One month ago we canceled Magic Brief and Hiro Analytics. This week, Asana. In both cases the tool that replaced them was built by someone on our team who has never written a line of code in her life. She used Claude, described what she needed, and had a working replacement in an afternoon.

There's a word for what she was doing. In February 2025, the computer scientist Andrej Karpathy posted something short on X that ended up naming a movement.

He was describing his own experience using Claude inside a coding editor, mostly as a curiosity. A year and a half later, it isn't a curiosity anymore. It's how a growing share of software gets built, and increasingly, it's being built by people who were never supposed to be able to build software at all.

Automation isn't new. Small businesses have been setting up workflows and triggers for years, if this then that, still leaning on a tool someone else built. What's different now is bigger than that. We're not configuring someone else's platform anymore, we're building our own. Call it a Claude artifact, call it whatever you want, what it actually is is software. SaaS, built by people with zero coding experience. At 58 Clicks, nobody on the team doing this is an engineer. They didn't build Asana. In less than a week, they built what replaced it.

The list, so far: a task manager that mirrors exactly how our team works, built to replace Asana. A dashboard that pulls Shopify, Klaviyo, Meta, and Google into a single view, doing the job Whatagraph, Hiro Analytics and Magic Brief used to do. A skill that reads and prioritizes the inbox, taking over what we used to pay Fixer for. Images, once downloaded from Envato Elements, are now created in our own image generating tool.

None of it required a developer. That's the part that would have sounded absurd two years ago, and doesn't anymore.

What this means for your Shopify app bill

The same shift is starting to show up somewhere our clients will feel directly: the Shopify app bill. Not the apps doing real work, Klaviyo, Gorgias, Skio for subscriptions, those run genuine infrastructure and aren't going anywhere. Yet. It's the smaller ones. A free shipping bar. A personalization widget. A countdown counter. Five dollars here, fifteen there, charged every month for something that, underneath, is a few lines of code. On a store running fifteen of these, the bill adds up quietly, and almost nobody stops to audit it.

Eventually, almost anything can be rebuilt this way, and that's exactly the problem. Taken too far, it turns into a distraction. We're not in the business of building apps, we're running an agency. Merchants aren't in the business of building apps either, they're selling product. So the question isn't whether something can be replaced. It's whether it's worth the time to replace it.

Building your own comes with a real trade off. On one side, it's fully personalized, built for exactly how you work, not the other way around, and it costs a fraction of what you were paying. On the other, someone has to maintain it, host it, and fix it when it breaks, and that someone is now you, not a vendor with a support line. For a five dollar shipping bar, that trade is easy. For anything with real complexity behind it, it isn't automatic.

Here's where to start. List every app on the store, sorted by price. Circle the ones doing something simple enough to rebuild. Multiply by twelve, not one, that's the number that makes the decision obvious.

That's the audit. There's a second move that has nothing to do with existing subscriptions: find the thing nobody's solved yet. The stockout nobody flags until a customer complains. The delivery route you need to work on every day. Those don't show up on an app bill, they show up as time nobody's counting. That's usually the bigger number.

Here's a candidate that's on our own list next: Calendly costs nineteen dollars a month, per seat. Claude, we already pay for. So the real comparison isn't twenty dollars against nineteen, it's nineteen dollars a month, per person, against zero marginal cost, because the tool doing the building is already on the bill. We haven't built our own scheduling tool yet, it's still an idea, but that's exactly why it's worth trying. Calendly is a multi billion dollar company, and what it actually does can be explained in an eight minute video.

That's the actual case for leveraging AI inside a company like ours, and it has nothing to do with hype. It's cost reduction. Every subscription on that list was a fixed cost that only went up as we grew. What replaced them is closer to a fixed cost that keeps absorbing more of the stack, the more we learn to use it.

This moves fast enough that half of what's in this email will read as basic by next summer. The upside is who gets to play now. Most of us have spent years being the ones who request software, not the ones who build it. Being on the other side of that for once, watching something you needed just exist because you described it clearly, feels genuinely amazing. It's a vibe, no other word for it.

We're planning Claude plus Shopify AI workshops for exactly this. Reply if you'd like to join.