Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next

Boris Cherny, creator of Claude Code, discusses how AI agents have fundamentally shifted software development from manual coding to an orchestration of autonomous, loop-driven processes. He argues that coding, as a technical task, is effectively solved for many use cases, shifting the focus toward domain expertise, cross-disciplinary collaboration, and "agentic" workflows.

Chapters

Chapter 1: Introduction

• Introduction of Boris Cherny, the creator of Claude Code.
• Discussion on the rapid evolution of modern software development tools.

Chapter 2: Claude Code Crowd Check

• Casual survey of the audience's preferred development tools (CLI vs. IDE).
• Background on Boris’s history as an engineer and author of TypeScript programming resources.

Chapter 3: Origin Story of Claude Code

• Developed as an innovation lab experiment at Anthropic to address "product overhang"—the gap between model capabilities and available tools.
• Initial team disbanded after creating Claude Code, MCP (Model Context Protocol), and the desktop app, before regrouping for further development.

Chapter 4: From Typeahead to Agents

• Transition from simple "typeahead" code completion to full-agent code generation.
• Early struggles with product-market fit were solved by aligning release cycles with more intelligent model iterations (Sonnet 3.5 to Opus 4.7).

Chapter 5: Is Coding Solved

• Cherny asserts that for his personal workflow, coding is 100% "solved" by AI.
• While AI handles standard stacks like TypeScript and React easily, complex, legacy, or niche codebases still pose challenges that require waiting for future model updates.

Chapter 6: Boris Personal Workflow

• Cherny manages hundreds of agents simultaneously, often directly from his phone.
• Highlights the power of "loops" (automated, recurring tasks) for managing PRs, fixing CI, and clustering feedback.

Chapter 7: Future Teams and Generalists

• Predicts a rise in cross-disciplinary generalists who combine engineering with design or data science.
• Notes that at Anthropic, even non-engineering staff use code to solve problems, democratizing technical contribution.

Chapter 8: SaaS Apocalypse Predictions

• Discusses how AI lowers the cost of software production, potentially devaluing certain "process-heavy" SaaS modes.
• Predicts a 10x increase in startups as small, AI-native teams gain the ability to compete head-to-head with large, slower incumbents.

Chapter 9: Audience Q&A Deep Dive

• Emphasizes that building "something people love" remains more important than the specific technology stack.
• Compares the democratization of coding to the printing press, predicting software will become a ubiquitous skill rather than a professional gatekept field.

Chapter 10: Closing and Whats Next

• Teases upcoming improvements to Claude Code and the ongoing development of massively parallel agent workflows.
• Reiterates that computer use and agentic delegation are the primary focus areas for the near future.