AI-powered coding services are steadily gaining traction, and Google perceives a significant challenge in the market leader, Anthropic’s Claude Code, which necessitates a direct countermeasure. The search behemoth plans to enhance its proprietary AI coding tools, as reported by The Information.
Google has decided to establish a dedicated “task force” comprising both researchers and engineers, whose sole mission will be to develop superior AI models specifically designed for code generation. This division will be spearheaded by Sebastian Borgeaud, a research engineer from Google DeepMind who previously oversaw foundational pre-training efforts. The initiative will also involve Google DeepMind’s CTO, Koray Kavukcuoglu, and Google co-founder, Sergey Brin. The internal consensus is that Anthropic’s latest Claude Code iterations currently outperform the Gemini equivalents, a situation the company is determined to rectify.
Google intends to focus on engineering AI models for internal utilization while simultaneously broadening the capabilities of its existing agents. The AI models earmarked for internal deployment will be trained using Google’s own extensive codebase, a move expected to boost employee efficiency; however, this proprietary nature will preclude them from being released publicly.
Currently, AI accounts for up to fifty percent of the code written in Google projects, and the company’s ambition is to increase this proportion substantially. They will be monitoring the adoption rate of their in-house coding mechanisms, known as Jetski, aiming to identify which personnel utilize these tools most frequently, organize necessary training sessions, and establish guidelines for their application in specific development scenarios. A spokesperson for Google confirmed the tools are already experiencing “tremendous uptake,” adding, “Their deployment is significantly accelerating the development of our AI models and tools—this is a core focus area for us.”
About the Author
