Discussion about this post

User's avatar
Markus Schmidberger's avatar

Amazing read, Ruben. There’s no doubt the 'Why' behind the AGI shift is compelling, and the 'What' (AI eating application software) is already underway.

But there is a massive, expensive blind spot in the current VC narrative: the 'How'

In the rush to celebrate 'vibe coding' and tools like Lovable, we’re witnessing a dangerous trend where VCs mistake a weekend demo for a production-ready product. As I explored in my recent post 'Beyond Vibe Coding', there is a massive iceberg beneath that 'magic' prompt.

https://builder.aws.com/content/3AboHhRNHxBSHIJvXhfOdMT0NSt/beyond-vibe-coding-why-real-ai-products-need-real-engineering

It’s easy to vibe-code a landing page. It is incredibly hard to engineer a real-time, bidirectional voice system that handles sub-second latency, manages complex state machines, and doesn't collapse at 3:00 AM under actual load.

When the hype settles, we’ll realize that 'vibe coding' is actually just the new technical debt. The winners won't be the ones who prompted the fastest; they’ll be the ones who understood that even in an AI-native world, real engineering - infrastructure, testing, and protocol-level architecture - isn't optional. It’s the only thing that scales.

Stop falling in love with the demo. Start looking at the 28 backend services required to make that demo stay alive in the real world.

No posts

Ready for more?