Google CEO Sundar Pichai has revealed that more than three-quarters of all new code at the company is now generated by Artificial Intelligence (AI). The figure marks a rapid acceleration in Google’s internal transformation, jumping from 50% just six months ago to 75% today, signalling the beginning of the “agentic” era in software engineering. “We’ve been using AI to generate code internally at Google for a while. Today, 75% of all new code at Google is now AI-generated and approved by engineers, up from 50% last fall,” Pichai said in a blog post written as the company is hosting its annual Cloud Next 2026 conference in Las Vegas.According to Pichai, despite the high percentage of AI involvement, human engineers remain the ultimate gatekeepers. “Every line of that code is reviewed and approved by engineers,” Pichai stated, clarifying that AI serves as a high-velocity draft writer while humans maintain the role of quality controllers.
Beyond coding: The rise of AI agents
Google is now moving toward “agentic workflows,” where engineers no longer just prompt AI for snippets of code but instead orchestrate “digital task forces.”“We’re now shifting to truly agentic workflows. Our engineers are orchestrating fully autonomous digital task forces, firing off agents and accomplishing incredible things,” Pichai said, explaining with an example that “a particularly complex code migration done by agents and engineers working together was completed six times faster than was possible a year ago with engineers alone.”
Another key highlights of this new workflow include rapid prototyping wherein Google teams moved from a concept to a fully functional native Swift app for the Gemini macOS release in just a few days using the internal platform called “Antigravity”.
Google’s ‘Customer Zero’ strategy
Pichai emphasised that Google is utilising its own engineering floor as a laboratory, acting as “customer zero” for its Gemini Enterprise Agent Platform.“To be the best partner, we always want to be ‘customer zero’ for our own technologies,” Pichai noted. The goal is to battle-test these AI agents at “Google scale” before offering the same infrastructure to global cloud customers.
