In an era where multitasking is often seen as a challenge, Boris Cherny, head of Claude Code at Anthropic, has turned it into an art form. Managing multiple Claude AI instances simultaneously, he exemplifies a level of efficiency that few could achieve. Cherny operates five Claude instances in his terminal, along with another five to ten sessions on the claude.ai platform. His method resembles that of an air traffic controller, orchestrating numerous workflows with precision.
Claude Code, which started as a terminal-based prototype in late 2024, was not designed to be a conventional product with user-friendly interfaces. Instead, it was crafted to allow seamless interaction between developers and the AI from the command line, enabling fluid task transitions between local instances and web-based environments. This innovative design has reshaped how AI-assisted coding works, pushing the boundaries of developer engagement with artificial intelligence.
Cherny’s insights into his workflow went viral in January 2026, with a social media thread that garnered millions of views. The thread demonstrated how he efficiently assigns tasks across different Claude instances and dynamically shifts work contexts based on project needs. The impact of this method has been significant, with productivity per engineer at Anthropic soaring by nearly 70% after adopting Claude Code. This remarkable figure is especially striking considering that the company's workforce tripled during the same period.
The evolution of AI interactions, as described by Cherny in discussions at Lenny’s Podcast and the Sequoia AI Ascent event, highlights a shift in developer practices. Initially, the interaction model was straightforward: developers communicated with the AI, which responded and evolved tasks through direct feedback. This first phase is now giving way to what Cherny calls “autonomous loops.” In this second phase, developers set tasks that Claude can execute independently, checking in only when necessary.
Cherny's vision extends into a third phase, which he describes as a significant leap in AI capabilities. He mentions orchestrating “a few thousand” sub-agents overnight to tackle complex tasks. This level of automation and delegation enhances efficiency and allows developers to focus on higher-level decision-making rather than getting bogged down in the details of the coding process.
As Anthropic continues to refine Claude Code, the implications for the broader AI development community are substantial. The transition from manual prompting to automated loops signifies an evolution that could redefine productivity standards in software engineering. With tools that can self-manage and autonomously execute tasks, the coding landscape is poised for a transformation that could yield even greater efficiency gains across the industry.
The future of AI-assisted development looks promising, with innovations like Claude Code leading the way. As more developers adopt similar approaches, the potential for increased productivity and the reimagining of traditional workflows could usher in a new era in software development. Cherny’s pioneering methods indicate what is possible when AI is fully integrated into the coding process.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.