Google CEO Says AI Generates 75% of Company Code as DeepMind Expands

Google CEO Sundar Pichai revealed that artificial intelligence now generates approximately 75% of the company’s code, marking a significant milestone in AI adoption within one of the world’s largest technology companies. The disclosure comes as Google simultaneously launched major upgrades to its autonomous research capabilities through new Deep Research agents and prepares to begin human trials of AI-designed drugs through its DeepMind spinoff.

According to The Times of India, Pichai’s statement represents one of the most concrete examples of AI’s integration into software development at scale. The 75% figure suggests Google’s engineering teams have fundamentally restructured their workflows around AI assistance, potentially setting a precedent for the broader technology industry.

Deep Research Agents Launch with Enterprise Integration

Google on Monday unveiled substantial upgrades to its autonomous research agent capabilities, launching Deep Research and Deep Research Max agents that integrate web data with proprietary enterprise information through a single API call. The new agents, built on Google’s Gemini 3.1 Pro model, can produce native charts and infographics within research reports and connect to third-party data sources through the Model Context Protocol (MCP).

“We are launching two powerful updates to Deep Research in the Gemini API, now with better quality, MCP support, and native chart/infographics generation,” Pichai wrote on X. The release represents Google’s bid to position its AI infrastructure as the backbone for enterprise research workflows in finance, life sciences, and market intelligence.

According to VentureBeat, the upgrade marks an inflection point in building AI systems that can autonomously conduct exhaustive, multi-source research traditionally requiring hours or days of human analyst time. The agents can fuse open web data with proprietary enterprise information, addressing a critical need in industries where information accuracy carries high stakes.

AI Drug Discovery Advances to Human Trials

Isomorphic Labs, Google DeepMind’s biotech spinoff, is preparing to begin human trials of drugs designed using Nobel Prize-winning AI technology. The company uses DeepMind’s AlphaFold platform, which predicts protein structures, for drug discovery applications.

“We’re gearing up to go into the clinic,” Isomorphic Labs president Max Jaderberg said at WIRED Health in London, according to Wired. “It’s going to be a very exciting moment as we go into clinical trials and start seeing the efficacy of these molecules.”

The timeline represents a delay from original projections. CEO Demis Hassabis previously stated the company would have AI-designed drugs in clinical trials by the end of 2025. Isomorphic Labs was founded in 2021 as a spinoff from Alphabet’s AI research subsidiary, leveraging DeepMind’s AlphaFold breakthrough that solved the decades-old protein folding problem.

AlphaFold’s Evolution and Impact

DeepMind’s AlphaFold 2, presented in 2020, used deep-learning techniques to predict protein structures with unprecedented accuracy. The company released an open-source version in 2021, making the technology available to researchers worldwide. AlphaFold 3, released in 2024, expanded beyond modeling proteins in isolation to predicting interactions with DNA, RNA, and other molecules.

Built from 20 amino acids, proteins fold into three-dimensional structures that dictate their function. Researchers had attempted to predict these structures since the 1970s, but the astronomically high number of possible shapes made this a painstaking process until AlphaFold’s breakthrough.

Enterprise AI Adoption Accelerates

The 75% code generation figure at Google reflects broader trends in enterprise AI adoption. Major technology companies are increasingly integrating AI tools into core development processes, potentially reshaping software engineering practices across the industry.

Google’s approach combines internal AI development with external-facing products. The Deep Research agents target enterprise customers requiring sophisticated research capabilities, while the company’s internal use of AI for code generation demonstrates practical implementation at scale.

The Model Context Protocol integration allows the new agents to connect with arbitrary third-party data sources, addressing enterprise needs for comprehensive data analysis. This capability positions Google’s AI infrastructure to compete with specialized research platforms in finance, pharmaceuticals, and market intelligence sectors.

Political and Regulatory Considerations

Google DeepMind has engaged in discussions with UK government officials regarding AI projects. The Financial Times reported that Morgan McSweeney held talks with Google DeepMind over an AI project, though specific details remain limited due to subscription restrictions.

These discussions occur as governments worldwide grapple with AI regulation and oversight. The UK has positioned itself as a leader in AI governance, hosting international AI safety summits and establishing regulatory frameworks for AI development and deployment.

What This Means

Google’s disclosure of 75% AI-generated code represents a watershed moment for enterprise AI adoption, demonstrating that leading technology companies have moved beyond experimentation to fundamental workflow transformation. This level of integration suggests AI coding assistants have reached production-grade reliability and effectiveness.

The simultaneous launch of enhanced Deep Research capabilities and progression toward AI-designed drug trials illustrates Google’s multi-pronged AI strategy. The company is leveraging its AI research advances across internal operations, enterprise products, and scientific applications.

For the broader technology industry, Google’s code generation metrics may establish new benchmarks for AI productivity gains. Companies evaluating AI adoption can reference Google’s experience as evidence of AI’s practical value in software development workflows.

FAQ

How does Google’s 75% AI code generation compare to industry standards?
Google’s 75% figure appears significantly higher than publicly reported metrics from other major technology companies. Most organizations report AI assistance rates between 20-40% for coding tasks, making Google’s adoption level notably advanced.

What makes the new Deep Research agents different from existing AI research tools?
The Deep Research agents uniquely combine web data with proprietary enterprise information through a single API call, while generating native charts and infographics. The Model Context Protocol integration allows connection to arbitrary third-party data sources, expanding research capabilities beyond typical AI assistants.

When will Isomorphic Labs’ AI-designed drugs reach patients?
Isomorphic Labs is preparing to begin human trials but has not specified exact timelines. The company previously targeted clinical trials by end of 2025, though current statements suggest preparation phases are ongoing without confirmed start dates.