Gene Kim and Steve Yegge have released a new book on 'vibe coding,' a development approach centered on using AI agents to accelerate software engineering. The authors bring together Kim's expertise in DevOps and systemic organizational design with Yegge's extensive engineering background at Amazon, Google, and Sourcegraph.

The book introduces the FAAFO framework—Fast, Ambitious, Autonomous, Fun, and Optionality—to help developers manage AI-driven workflows. It emphasizes that while AI increases speed, the human developer remains fully responsible for the final product, including troubleshooting errors and managing system stability.

The authors address common risks such as AI agents deleting tests or creating unmodular code. They propose a structured approach involving three development cycles—inner, middle, and outer—to detect and correct errors. The text also details specific concepts like 'tracer bullets' and the Model Context Protocol (MCP).

The guide is intended for a broad audience, including experienced programmers and team leads. It encourages readers to start small with available tools, such as GigaCode or SourceCraft, and focus on building robust engineering processes rather than relying solely on the capabilities of a specific AI model.