It's so easy and fast to build stuff with AI 🪄
Well, so easy in fact, that you may end up searching for something new to implement, even if you probably don't need it 💁🏼♂️
Since AI is a gambling loop, you need to pay extra attention to this 👇
One personal project at the time
Back in the days, my log of "app ideas" was growing constantly, with the capacity of working on (maybe) one project at the time.
And that was indeed a good constraint:
Most of those "projects" were temporary/unnecessary 🙌
Coding required a lot of time. Doing one project meant leaving another behind. So you had to be careful, and choose only what mattered.
Guess that's not it anymore 💁🏼♂️
AI made coding trivial. You have more time and capacity. What do you do with it? 🤔
One more app
In my X feed (bubble), I see people posting about all sort of games, apps, libraries.
It makes sense, coding is fast and easy. But did you really need that? 🤔
If coding is easy now, the "hard" part moved somewhere else: make something people want, infrastructure, deployment, maintenance.
But, you can skip all these steps if you make a new app, right? And another, and another 😬
Easy, but not free
Coding is easy, but it still sucks your time.
Agents can build everything, but you still need to prompt, supervise, debug ⏱️
Coding with AI is so easy (and addictive), that you can easily skip the question "Do I really need this app?", and just go straight to the prompt.
And another prompt, and another, and another.
Before you realize it, another day is gone, with another half made app 😬
When coding was hard
"Suck" you time is the correct analogy.
Before AI, even if the answer to "Do I really need this app?" was no, there was still value in the project:
- Learn a new stack, library, skill, language
- Build your portfolio of projects to showcase
- Learn to focus, work and ship a complete project
With AI, best case you got to learn a couple of new tricks by reading the AI code (if you do read it).
Worst case, you just wrote (or spoke) a prompt, and watched. And time is gone 😬
You may not even need an app
Not only you may not need that app, but even assuming you do, you may not need for it to be a full custom-implemented app.
Ironically, AI itself can make most app ideas unnecessary 💁🏼♂️
I had an idea for a new language learning setup. AI feedback, history, exercises generation.
Turns out an app is not needed at all. I build everything with a local SQLite db, that the AI (Codex) uses as local storage.
I can practice without needing an app, just prompting 🪄
I have no way to confirm this, but it seems to me most people are jumping on whatever project just to "use AI more".
Races to whoever can finish the weekly limits the fastest, because not maxing your limit is a "waste". So let's start a new project 🙌
If this is you, pay attention. Don't sell all your time to AI, for no benefit.
See you next 👋