Google Cloud Introduces Three Levels of Agentic Software Development for Developers

Google Cloud has launched a new approach to agentic software coding with three distinct levels aimed at streamlining the development process. This development follows the success of Firebase Studio, a tool that helps developers create AI-powered applications through natural language prompts. The three levels, ‘Ask,’ ‘Agent,’ and ‘Agent (Auto-run),’ provide varying degrees of autonomy for Google’s AI service, Gemini. ‘Ask’ is a conversational mode suitable for brainstorming and problem-solving, while ‘Agent’ offers the ability to suggest changes with developer approval. The most advanced ‘Agent (Auto-run)’ mode can autonomously reason, apply changes, write tests, and refactor code, helping both new and experienced developers. Despite these innovations, Google Cloud reassures developers that their roles are not in danger, emphasizing that the tools are designed to support, not replace, human expertise. The company also introduced new tools, including Gemini CLI, for seamless integration into the developer’s workflow, enhancing productivity with AI assistance at every level. The evolution of these technologies is a step towards democratizing software development, allowing even non-technical staff to participate in the coding process.