Google Forms AI Coding Strike Team as Sergey Brin Pushes DeepMind to Pivot to Agents
Google has assembled a dedicated strike team focused on improving its AI coding models, as the company faces intensifying competitive pressure from OpenAI's Codex, Anthropic's Claude, and a rapidly growing field of AI coding tools. Simultaneously, Google co-founder Sergey Brin has directly urged DeepMind staff to aggressively pivot their work toward AI agents — systems that can autonomously complete multi-step tasks — signaling a top-level strategic push to close the perceived gap with competitors in agentic AI.
The Coding Strike Team
The dedicated coding model team is a response to mounting evidence that Google's AI coding capabilities lag behind the competition in developer adoption metrics. GitHub Copilot (powered by OpenAI), Cursor (using Claude), and Windsurf have captured significant developer mindshare, while Google's own coding tools — including Gemini Code Assist — have struggled to achieve comparable adoption despite Google's underlying model capabilities. The strike team approach suggests Google views coding as a priority battleground requiring focused engineering resources rather than incremental improvements to existing products.
Brin's Agent Push at DeepMind
Sergey Brin's direct intervention at DeepMind is notable for its urgency. According to sources, Brin told DeepMind researchers that the lab must "aggressively pivot" to AI agents to keep pace with competitors who are moving faster in agentic capabilities. DeepMind has historically focused on foundational research and game-playing AI, but the commercial AI race has created pressure for the lab to translate its research strengths into market-competitive products faster than its traditional research timeline allows.
The Competitive Context
Google's moves come as OpenAI's operator-mode agents and Anthropic's computer-use capabilities have demonstrated that agentic AI is no longer theoretical. Enterprises are beginning to deploy agents for real workflows, creating a window for whoever can deliver the most reliable and capable agent infrastructure to capture durable enterprise relationships. Google has the model capability but has been criticized for slow product iteration — the strike team and Brin's push appear designed to address exactly that.
The Bottom Line
Google is clearly in catch-up mode on coding and agents despite having world-class underlying AI research. The strike team and Brin's DeepMind intervention are the right moves, but execution will determine whether they translate into market share. Watch for accelerated Gemini Code Assist updates and new agentic product launches from Google in the second half of 2026.