Vibe Coding: The Infrastructure Problem
The vibe coding trend has gained significant attention lately, but I believe we need a reality check on what it can actually deliver for production applications.
My brief explorations of tools like Lovable.dev and Bolt.new revealed that they rely on traditionally-coded infrastructure like Supabase to function effectively. This suggests vibe coding's ultimate success depends on the quality and breadth of what lies beneath it.
For prototyping and simple applications, vibe coding shows promise. But for complex production systems? I remain skeptical. I'm curious about others' experiences using these tools for anything beyond basic implementations.
The path forward, as I see it, requires specialized infrastructure tools designed for specific domains. In my own work, I've developed the ChainReact.NET library with high-level abstractions that span the entire application stack (frontend, backend, business logic, etc.) for line-of-business applications. This approach could provide the foundation that vibe coding needs to become truly effective in that domain. One-size-fits-all solutions likely won't succeed. Instead, purpose-built infrastructure for different application categories offers more promise.
What do you think? How might we bridge the gap between vibe coding's lofty ambitions and production reality? Which infrastructure areas most urgently need development to make vibe coding viable?
What vibe coding changes is the prototyping phase. Going from 0->1 doesn’t take a team. You need the team, like you say, to get to the next level of scaling or complexity (at least for now)
We have to realize the next generation of programmers are 10 years old right now, and the next generation of business owners are also about that age. There will be a gap between technical expertise between the two, as usual, hence leading to business. Just like we saw, our business owners did not give two shits about how things work and why they break, so long as they work, and so long as things are fixed (however the heck). So, if the new generation of coders vibe code their ass off to make things and vibe code their ass off to fix things, that will be the new nature of programming work. The business has never cared for the most part.
Fixing things requires more finess than adding features. Because it needs to be fixed. Not fixed means you are out of business. Vibe coding is rolling a dice to see if something works. The question is whether these systems can stack that dice to converge to self healing within the time needed.
Maybe they will maybe they won't, or maybe it depends on the problem.
I think you need a expert human in the loop. Maybe fewer of them for a given weight of functionality.