In this episode of The Vergecast, host David Pierce explores how Anthropic’s Claude Code is revolutionizing software development and everyday digital tasks by automating coding and information management, featuring insights from creator Boris Cherny on its evolution and broader applications. The discussion also highlights the importance of privacy and safety when using AI tools, as well as the need for user vigilance amid rapidly changing technology.
On this episode of The Vergecast, host David Pierce dives into the world of AI-powered coding, focusing on Anthropic’s Claude Code and its impact on software development. David opens by reflecting on his own chaotic information habits and the catharsis of reading books, before pivoting to the main topic: how AI tools like Claude Code are transforming the way people write and interact with code. He shares a personal example of using Claude Code to organize his scattered notes across multiple apps, highlighting how AI can automate tedious digital tasks and make information management more efficient.
The centerpiece of the episode is an in-depth interview with Boris Cherny, the creator of Claude Code at Anthropic. Boris discusses the evolution of Claude Code from a developer tool to a product with much broader appeal, noting that it now writes nearly all of his code. He explains how improvements in Anthropic’s models, especially with the release of Opus 4.5, led to a sudden leap in code quality and reliability, allowing him to fully delegate coding tasks to the AI. Boris also describes the shift in his own identity as a coder, likening the transition to moving from playing an instrument to conducting an orchestra—he now directs and reviews rather than writes code line by line.
Boris and David explore the surprising ways non-developers have adopted Claude Code and its offshoot, Co-Work, for tasks beyond traditional programming. From data analysis to organizing screenshots and even paying parking tickets, users are finding creative applications for AI agents. Boris emphasizes that while Claude Code started as a terminal-based tool for engineers, the team quickly realized the need for more accessible interfaces for non-technical users. This led to the development of Co-Work, which wraps Claude Code’s capabilities in a safer, more user-friendly environment, complete with safeguards like virtual machines and deletion protection.
The conversation also addresses the risks and responsibilities of giving AI agents access to personal data and computers. Boris outlines Anthropic’s commitment to safety, privacy, and security, especially for enterprise customers, and details the technical and policy safeguards in place. However, he acknowledges that challenges like prompt injection and evolving terms of service remain, and users should be thoughtful about what data they share with AI tools. This theme is further explored in a segment with Verge AI reporter Hayden Field, who discusses the complexities of data privacy, the shifting landscape of AI company policies, and the importance of informed consent when connecting sensitive accounts like Gmail to AI services.
The episode wraps up with a listener question about whether to upgrade a smartphone now due to rising RAM prices, with the consensus being that most users can safely wait unless they’re already due for an upgrade. Throughout, the hosts emphasize the rapid pace of change in AI, the need for ongoing vigilance around privacy, and the exciting, sometimes unpredictable ways that AI is reshaping both technical and everyday tasks.