Keep pulling the thread on Guillermo Rauch.
The Next.js framework, created by Vercel, powers applications for companies including Claude.ai and Walmart.
The cost of creating software is fundamentally approaching zero.
An AI agent, Claude Bot, was used to deploy 91 separate applications to Vercel within a single week.
As AI makes software deployment ubiquitous, the key differentiating value for companies is shifting from velocity to taste and curation.
In an AI-driven world, the value of brand and trust as business moats has increased dramatically.
The primary customer for infrastructure companies is shifting from human developers to autonomous AI agents.
Vercel is developing a command-line subcommand that will allow AI agents to make infrastructure purchasing decisions, such as buying a database.
Vercel's bet on open source inadvertently created a competitive advantage, as its open-source projects became the training data for AI agents.
A company with an open-source project has implemented a rule that pull requests must contain no code, only requirements and tests, with an AI model handling the implementation.
Vercel's internal principle for AI development is to avoid 'random acts of AI' and not superficially add AI features to products.
Companies will need to shift from 'developer evangelism' to 'agent evangelism' to make their products compelling to AI agents.
LLMs prefer training data with no noise, HTML tags, or styling, seeking only the pure signal of the content.