Boris Cherny of Anthropic highlighted that AI has effectively solved coding for many use cases, enabling fully autonomous code generation and workflow automation through AI agents, which frees developers to focus on higher-level tasks. He envisions a future where software development becomes democratized and ubiquitous across disciplines, fostering innovation and competition in startups while AI tools become increasingly intuitive and integrated within existing ecosystems.
Boris Cherny, the creator of Claude Code at Anthropic, shared insights into the evolution and future of software development powered by AI. He recounted how Claude Code began as an experimental project within Anthropic Labs in late 2024, initially struggling to gain traction due to limitations in early AI models. However, with successive model improvements, particularly starting with Opus 4 in May, Claude Code experienced exponential growth, eventually reaching a point where the AI writes 100% of the code for Boris himself. This marked a significant milestone, leading him to declare that coding, as a task, is effectively solved for many use cases.
Boris described his personal coding setup, which heavily relies on AI agents running on his phone and desktop, managing hundreds of sub-agents and loops that automate tasks such as fixing CI issues, managing pull requests, and gathering feedback. He emphasized the power of “loops,” a feature that schedules repetitive jobs for AI agents, allowing continuous and autonomous work even when his laptop is closed. This approach exemplifies how AI can handle complex, parallelizable workflows, freeing human developers to focus on higher-level tasks.
Looking ahead, Boris predicted a shift in team dynamics and skill sets, with a rise in cross-disciplinary generalists who combine engineering with design, product management, data science, and other fields. He noted that at Anthropic, everyone on the Claude Code team, including non-engineers, writes code, reflecting a democratization of software development skills. This trend aligns with his broader vision that coding will become as ubiquitous and accessible as reading and writing, enabling domain experts like accountants to build software tailored to their needs without deep programming expertise.
On the broader impact of AI on software products and startups, Boris argued against the notion of a “SaaS apocalypse.” Instead, he highlighted that while AI reduces barriers like switching costs and process complexity, fundamental business advantages such as network effects and scale economies remain vital. Moreover, AI lowers the entry threshold for startups, enabling small teams to compete with large incumbents by building AI-native products from the ground up, which could lead to a surge in innovation and disruption in the coming decade.
Finally, Boris addressed questions about the future of AI tooling, including the balance between cloud-based and local AI models, the integration of AI with existing software ecosystems via MCP connectors, and ongoing product innovations like Claude Design and enhanced multi-agent coordination. He stressed that as models improve, much of the complexity in managing AI workflows will shift from users to the AI itself, making these tools more intuitive and powerful. Overall, Boris painted an optimistic picture of AI-driven software development becoming more efficient, accessible, and transformative across industries.