Vibe Coding and Getting Rich Quick
Is Everyone Building the Same Thing?
The rise of vibe coding apps like Lovable and Replit has been notable. Beyond raising huge funding rounds, there has been a narrative tsunami around these apps.
The basic idea is that anyone can use these apps to build any type of software. From mobile apps to B2B SaaS companies, we're in an era where you can, theoretically, build an app with a prompt and a dream.
But what does this actually look like in practice? I'd suspect that the results are much more mixed.
For starters, it's unclear how many complex apps are actually in production. Some public reports suggest that Lovable users are building more than 25,000 apps every single day. That said, there aren't any official figures on the number of Lovable apps in production. Replit's CEO recently said that there were over 100,000 Replit-built apps in production. However, the consensus is that it is good for shipping MVPs and simple apps.
Also, many people are making the same thing. The stereotype is the calorie-counting app. With just a few prompts, you can build a calorie-counting mobile app. The initial movers have made hundreds of thousands of dollars on these apps (perhaps more), but there has been a flood of copycats. Notwithstanding this fact, the same ideas re-emerge in the desire to get rich quickly.
Then there is the durability of the code. Yes, vibe coding apps can get your code to production. But what happens if your app goes viral? What happens if the app can't handle a burst of incoming traffic? There are no easy answers for people who have quickly vibe-coded a home run of an app idea.
And then there is security. A common joke on Twitter is to try searching for "OPENAI_API_KEY=" on GitHub. There, you can see vibe coders who have inadvertently pushed their OpenAI API key, which can give anyone with the key the power to use those API keys in their own projects.
To be clear, I think the intentions behind these vibe coding apps are great. While the promise of AI may be overhyped in some areas, it is perhaps underhyped in software. I can speak from firsthand experience. I've avoided tens (or hundreds) of hours on Google and Stack Overflow by chatting with LLMs. It certainly doesn't get everything right, but it is an accelerant as I try to figure out the right answer.
AI is democratizing development, but it is still a long way away. Like I've said before, I think the best approach is to work on pressing B2B problems in niche sectors and industries. Instead of deferring all development decisions to vibe coding apps, it's best to intelligently develop with LLMs. It's more challenging than simply pressing "Accept All," yet it is much more rewarding in the long run.
The gold rush continues, yet the long-term benefits will accrue to those who develop good habits with these amazing tools. I'm not saying I'm doing things perfectly, but I'm trying to avoid flashy object syndrome as I work on my product. As always, we'll see what happens!
Prompt of the Week
I typically use LLMs to shore up my weaknesses. It's easy to think of attributes or traits we need to improve, whether that is in our personal or professional lives. Ultimately, it's easy to take our strengths for granted.
I've been trying to reverse this thought by deliberately doubling down on my strengths. While I think I have a good baseline of what my strengths are, I decided to ask ChatGPT. Try it for yourself!
"Considering what you know about me and my goals, what are my biggest strengths that I'm not taking advantage of? What should I be doubling down on? Give me a 30-day framework on how I can leverage my strengths to achieve my goals."