The Temptation to Build Everything

There’s this weird feeling when adopting new tools as a software engineer. If you’re even slightly AI pilled you’ll almost always have these questions:

• Could this be a prompt? (or a skill)
• Can I just build this myself?

In my opinion this happens for two reasons:

1. A deep understanding of the friction surrounding bringing more tools in your life.
2. The fun associated with building and starting something new.

Adopting new tools sucks

Let’s say you’ve gone through the painstaking process of prompting your agent to find the best tool for a specific job. It comes back with a shortlist of tools you’ve never heard of, you just click the first link you see. You create an account, you go through onboarding, you’re tired. There’s a new dashboard to learn, menus to navigate. You finally find the feature you needed and boom, paywall. This caveat was definitely buried somewhere in your agent’s response that you didn’t read.

Paying small subscription fees for tools that made your life even marginally better made sense 24 months ago. Building and deploying software had orders of magnitude more friction than paying $3.99 a month. But now you’re at most half a dozen prompts away from something that does what you need. So screw this you think, I’ll just build it myself.

Building software is more fun than using it

Ah the smell of a new git repo, domain, crap Claude-ish logo. It’s addictive. And there you are 20 minutes later with a live version of the thing you need. No paywall, no useless features in your way and it was quicker than figuring out the pre-existing options.

I personally think this is great. Instead of spending time doing something tedious I get to do what I enjoy. The arguments of the naysayers of this approach are also getting extremely thin:

• “It won’t scale” - I’m the only user
• “It won’t be secure” - Try this: “Hey Claude, make sure the app is secure”. Also you aren’t a bank, no one cares about this data anyways.
• “You’ll spend so much time maintaining it” - unlikely to be more than 2 prompts per month.

Honestly though, none of this was ever really about saving $3.99 a month. That’s just the excuse. You’re building it because it’s fun, and there’s nothing wrong with that as long as you admit it to yourself.

When to just pay the $3.99

There’s some fine print here though. All of the above only holds while you’re the only one who cares. The moment a teammate starts depending on your tool those three arguments come back, and this time they’re right.

It also adds up. 2 prompts per month is nothing for one app, but the temptation is to build everything, and a dozen small apps is a dozen things that can break and a dozen places your data now lives. And since building is the fun part, you’re the worst possible judge of whether something is worth building. It will always feel like a good use of your afternoon.

My rule of thumb:

• Build it if you’re the only user, it breaking is annoying rather than expensive, and you’d happily delete it in 6 months.
• Pay for it if other people rely on it or losing the data would really hurt. Or the value isn’t in the code at all (proprietary data, integrations, etc.).

Yes, a password manager is “just” a CRUD app. Please don’t build your own password manager.