When Custom Code Is Worth Eight Years, and When It Isn't

A few days ago someone posted Homegames to Hacker News. An open-source game platform, built solo, for eight years. Ships to real users. I love reading about projects like that.

But it also made me think about the founders I talk to every week. Someone opening a Pilates studio in Gangnam. Someone launching a small SaaS. Someone running a two-person marketing agency who needs a client portal. None of them have eight years. They have this month.

The question worth asking: when does your project actually deserve custom code, and when should you just ship it?

The scenario I keep seeing

It's always some variation of the same thing. You need a tool. Logins for a few users. A database that stores stuff. Emails that go out when something happens. A page or two of UI. Payments if you're taking money.

For example: a booking tool for your studio. A client portal for your agency. An internal request tracker for your team. A private community for your customers.

Traditional path: pick a framework, set up auth, wire in a database, add a payments gateway, configure email, buy a domain, set up hosting, glue it all together, style it, ship it. If you're a solo dev who knows what you're doing, six weekends. If you're outsourcing to an agency, three months and six figures.

Down the no-code path

You describe what you need. The platform gives you logins, a database, deploy, notifications, a live URL, and SSL by default. You don't touch infrastructure. On DontCode, the AI is fine-tuned specifically for app-building work, so you get less of the generic "let me generate a landing page for you" behavior and more of the "here's your working feature" behavior.

Your time goes to the parts that actually make your app different: how it feels for your customer, what data you collect, what emails you send, what checkout looks like.

When custom code is still worth it

Building a game engine. A novel database. Consumer software where a thousand tiny UX details decide who wins. Anything where the code itself is the product.

Most business software isn't that. Most business software is a form, a database, some logic, and an email. That doesn't deserve eight years. It doesn't even deserve six weekends.

The math to run

Take the thing you're about to build. Ask: is the value in the technical implementation, or in the business logic, customers, content, workflow?

If it's the second one, spending months on infrastructure is a rounding error compared to your competitors. They'll ship first.

If you want to see the shortcut version, poke around DontCode and see what ships built-in. Your MVP might be a lot smaller than you think. And if you want to read more comparisons like this one, the blog has a growing shelf of them.