Google co-founder Sergey Brin has acknowledged that the company is lagging behind Anthropic in the development of AI coding agents, according to a memo reported by The Information. In the internal communication to DeepMind employees, Brin stated that “every Gemini engineer must be forced to employ internal agents for complex, multistep tasks.” This directive underscores Google’s urgency to close the gap with Anthropic, whose tools have emerged as leaders in the AI coding space.
The memo, dated April 20, 2026, reflects growing concern within Google’s AI division about Anthropic’s advancements in agent-based systems capable of automating software development workflows. Brin framed the initiative as a critical step toward building AI systems that can improve themselves—a long-term goal in the pursuit of artificial general intelligence.
In response, Google DeepMind has assembled a specialized strike team tasked with accelerating progress in AI coding models. Led by Brin and DeepMind’s Chief Technology Officer, the team’s primary objective is to enhance Gemini’s capabilities in handling complex, multi-step coding challenges through internal agent usage. The move signals a strategic pivot toward internal tooling as a competitive necessity.
Anthropic’s Claude 3 model family, particularly its Opus variant, has demonstrated strong performance in coding benchmarks such as HumanEval and MBPP, where it consistently outperforms earlier versions of Google’s Gemini models. These results have contributed to the perception that Anthropic is currently ahead in the race to deploy practical AI coding assistants.
The strike team’s focus aligns with broader industry trends where major tech firms are investing heavily in agent-based AI systems. Unlike traditional language models that generate code snippets in response to prompts, AI coding agents are designed to plan, execute and debug multi-stage programming tasks autonomously—mirroring the workflow of a human software engineer.
Google has not disclosed the exact size or timeline for the strike team’s deliverables, but internal sources indicate that the initiative is being treated as a high-priority effort within DeepMind. The company has previously invested in AI-assisted coding through tools like AlphaCode and integrations with its Vertex AI platform, though none have yet matched the agent-centric approach pioneered by Anthropic.
This development comes amid intensifying competition in the generative AI landscape, where companies are racing to differentiate their offerings beyond basic chatbot functionality. AI coding agents represent one of the most promising applications due to their potential to significantly increase developer productivity and reduce software development cycles.
As of April 21, 2026, neither Google nor Anthropic has released public benchmarks comparing their latest agent-based systems directly. However, third-party evaluations from platforms like Hugging Face and LMSYS Org continue to show Anthropic’s Claude models maintaining an edge in reasoning and code generation tasks.
The outcome of this internal push could influence Google’s positioning in the enterprise AI market, where demand for reliable, secure, and intelligent coding assistance is growing rapidly. Companies are increasingly evaluating AI agents not just as experimental tools, but as potential fixtures in DevOps pipelines.
For now, the memo and subsequent formation of the strike team confirm that Sergey Brin views catching up to Anthropic as an immediate priority for Google’s AI ambitions. Whether this effort will yield measurable progress in the coming months remains to be seen, but it marks a clear acknowledgment of competitive pressure in one of AI’s most consequential application domains.