# vibecoding
Vibecoding and the second wave of AI coding
The AI coding category went from Karpathy's offhand essay to a $30B market in eighteen months. The first wave aimed at developers and hobbyists. The second wave is aimed at non-technical professionals with production needs, and the infrastructure for it just shipped.
> essays in this thread
The vibecoding bet: why the next ten million developers will never write code
63 percent of vibecoding users identify as non-developers. The market is being treated as a developer-tools category, and the largest underserved segment is being missed by every well-funded player. The contrarian thesis for who actually owns the second wave.
Launching AIgateway: 150 AI Models Behind One Key
Every AI startup rebuilds the same plumbing. OpenAI here, Anthropic there, Gemini, Groq, Together, Replicate, each with its own auth, retries, streaming, rate limits. I paid that integration tax 30 times. So I built AIgateway: one API, one key, one bill, 150+ models, edge-native on Cloudflare. Free Kimi K2.6 until Apr 30.
Your AI Wrapper Has No Moat. Here's How to Build One
The problem with 'AI wrapper' startups and how to build defensible products on top of foundation models. The 5 layers of AI product moats, with real examples, a self-assessment framework, and the playbook for each layer.
Latency is the New Downtime: Why Speed is Your AI Product's Moat
In the GenAI era, slow is the same as broken. A data-driven guide to inference speed, latency budgets, and why edge compute is a survival mechanism, not an optimization.
The Search Bar is Dying: How AI is Replacing Navigation with Intent
We're moving from navigation (clicking menus) to intention (typing what we want). A deep analysis of why your product design is about to become obsolete, with examples, data, and a redesign framework.
> next reads