The Aboard Podcast Of Course There’s an AI Bubble
Oct 21, 2025
The discussion dives into the plummeting traffic of vibe-coding tools and the looming threat of an AI market crash. Experts analyze the structural vulnerabilities of startups that repurpose large LLM outputs and the challenges AI faces in finalizing production-quality software. They predict a shift toward domain-specific AI tools, moving away from universal applications. As they explore the potential consequences of a market bubble, they stress the importance of building resilient AI companies focused on real value.
AI Snips
Chapters
Books
Transcript
Episode notes
Lovable's Viral Rise And Fall
- Paul shares Chris Pisarski's viral LinkedIn post asking, "Is lovable dying?" which launched the discussion on vibe-coding popularity.
- That example anchors their observation that web traffic and excitement around vibe-coding tools fell after a summer peak.
First Mile Magic, Last Mile Problem
- Vibe-coding tools excel at the first mile by filling blank pages and generating prototypes quickly.
- They struggle with the last mile because finishing software requires deep, embedded human detail and product judgment.
Code Without Intent
- LLMs generate code as expression without purpose and therefore make assumptions that break real products.
- That lack of intentionality creates long, liability-prone code that misses domain-specific needs like authentication flows.
