- Google is reportedly creating an AI ‘strike team’ to challenge Claude Code
- The company wants to build AI agents that can assist coding and research
- Yet the result might never be released to the public
One of the most significant ways that AI is changing the world is through the rise of AI agents that can write code. Instead of spending hours crafting and debugging code, programmers are now able to simply ask an AI model to get to work for them.
Google’s Gemini AI is pretty adept at this, but that’s not good enough for the company, which has reportedly been so rattled by the progress of rival Anthropic and its Claude Code AI agent that Google has formed a “strike team” to address the problem.
Indeed, according to a report from The Information, Google co-founder Sergey Brin has warned that Google can’t afford to waste any time. Writing in a memo to the company’s AI engineers and researchers, Brin said: “To win the final sprint, we must urgently bridge the gap in agentic execution and turn our models into primary developers” of final code.
And this isn’t just some hands-off effort from the Google executive team. Both Brin and Google DeepMind Chief Technology Officer Koray Kavukcuoglu have been “directly involved with the strike team, showing its importance to Google’s top leadership,” The Information claims. That aptly illustrates how seriously the company appears to be taking the threat from Anthropic.
Will it ever see the light of day?
What is Google aiming for here? Well, the most obvious answer would be to maintain a competitive edge against a popular rival in an emerging field that stands to make dominant companies a whole boatload of cash.
But there’s more to it than that. As The Information noted, “The end goal is Al takeoff, or Al that can improve itself, which has been a particular focus for Google co-founder Sergey Brin.”
Indeed, the report puts it this way: “An advanced coding agent, combined with Al that can solve math problems and run experiments, could one day automate the work of Al researchers and engineers at a large scale.” That’s something that another rival, OpenAI, has already developed for its own researchers working on internal AI projects, with the company also building a GitHub rival to help streamline its employees’ work.
The priority appears to be to improve Gemini’s ability to work with Google’s own code. That would improve its performance when used by Google’s engineers and researchers for their internal projects, with less of the emphasis being placed on coding tools for everyday coders who sit outside the organization. Given that an improved AI agent will have been trained on Google’s code, it couldn’t be released publicly.
So before you get your hopes up, you might not see the work of Gemini’s strike team, at least not initially. But as this report points out, the improved internal tools “could theoretically help create better models that could then see a public release.”
For now, all we can do is wait and see whether Google is able to close the gap between its own AI agents and Anthropic’s Claude before the gulf becomes insurmountable.
