In a significant shift in the tech industry, Vercel’s CEO Guillermo Rauch is spotlighting a new approach to software development. Instead of designing just for human users, the future is set to serve AI agents—tools that developers, and even non-developers, will soon rely on for their tasks.
This evolution means that coding isn’t just for people anymore. Developers are now writing code that AI agents can read, interpret, and build upon. On the ‘Sequoia Capital’ podcast, Rauch explained how large language models (LLMs) are reshaping software architecture, signalling that our runtimes, languages, type checkers, and frameworks must evolve to support AI-driven workflows.
Vercel, a web infrastructure startup valued at £3.25 billion, recently raised $250 million in Series E funding, supported by well-known investors like Accel, Tiger Global, and GV. Analysts suggest that while websites and apps will continue to exist, direct human interaction may give way to AI assistants that aggregate and streamline information, fundamentally altering our online experience.
Of course, this shift isn’t without its challenges. If you’ve ever wrestled with tricky system errors, you’ll appreciate the caution advised by companies like Patronus AI. They warn that even a small 1% error rate per step can accumulate quickly—by the 100th step, errors could soar to 63%. Thankfully, practical measures such as filters and more robust guidelines can help control these risks as the industry adapts.
