Google CEO Says AI Generates 75% of Company Code as DeepMind Drugs Enter Trials

Google CEO Sundar Pichai revealed that AI now generates 75% of the company’s code, marking a dramatic shift in how the tech giant develops software. The disclosure came as Google launched new Deep Research agents and its DeepMind spinoff Isomorphic Labs prepares to begin human trials of AI-designed drugs.

AI Code Generation Reaches 75% at Google

Pichai’s announcement highlights Google’s aggressive adoption of AI-powered development tools across its engineering teams. According to The Times of India, this represents a significant acceleration from previous AI coding adoption rates at major tech companies.

The 75% figure suggests Google’s engineers are using AI assistants for routine coding tasks, debugging, and potentially more complex programming challenges. This level of AI integration in software development indicates how quickly generative AI has moved from experimental tool to core productivity driver.

Google’s internal use of AI for coding likely leverages its own models and tools, providing real-world testing for AI capabilities before releasing them to external developers. The company has not disclosed which specific AI models power this internal code generation.

Deep Research Agents Launch with Enterprise Focus

Google on Monday unveiled Deep Research and Deep Research Max agents, built on the Gemini 3.1 Pro model. According to VentureBeat, these agents can “fuse open web data with proprietary enterprise information through a single API call.”

The new agents introduce several enterprise-focused capabilities:

• Native chart and infographic generation within research reports
• Connection to third-party data sources through Model Context Protocol (MCP)
• Integration of public web data with private enterprise information
• API-based deployment for developer integration

Sundar Pichai announced on X that the agents offer “better quality, MCP support, and native chart/infographics generation.” The release targets finance, life sciences, and market intelligence industries where research accuracy is critical.

Enterprise Research Automation

The Deep Research agents represent Google’s attempt to automate multi-source research that traditionally required hours or days of human analyst work. The ability to combine proprietary enterprise data with public information addresses a key limitation of previous AI research tools.

However, the agents are currently available only through the Gemini API, not in the consumer Gemini app. Several users on social media noted this limitation, with some questioning why Google restricts the feature from paying Gemini Pro subscribers.

Isomorphic Labs Advances AI Drug Discovery

Isomorphic Labs, Google DeepMind’s biotech spinoff, will soon begin human trials of drugs designed using AI technology. According to Wired, company president Max Jaderberg confirmed the clinical trials are approaching: “We’re gearing up to go into the clinic.”

The trials will test drugs created using DeepMind’s AlphaFold platform, which predicts protein structures using deep learning techniques. Isomorphic Labs was founded in 2021 specifically to apply AlphaFold’s protein folding predictions to drug discovery.

This timeline represents a delay from previous projections. CEO Demis Hassabis had previously stated the company would have AI-designed drugs in clinical trials by the end of 2025. The upcoming trials will provide the first real-world test of whether AlphaFold’s protein predictions translate into effective therapeutics.

AlphaFold’s Drug Discovery Evolution

AlphaFold revolutionized protein structure prediction starting in 2020, solving a decades-old challenge in computational biology. The platform moved beyond isolated protein modeling with AlphaFold 3 in 2024, which can predict interactions between proteins, DNA, and RNA.

The transition from research tool to drug development platform represents a critical test for AI’s potential in pharmaceutical discovery. Traditional drug development takes 10-15 years and costs billions of dollars, with high failure rates in clinical trials.

Google’s Broader AI Integration Strategy

The 75% code generation figure, new research agents, and advancing drug trials illustrate Google’s comprehensive AI integration across multiple business areas. The company is simultaneously using AI internally for software development while building external products for enterprise research and pharmaceutical discovery.

This multi-pronged approach positions Google to capture value from AI across different market segments. Internal productivity gains from AI coding tools can reduce development costs while external AI products generate new revenue streams.

The strategy also provides Google with extensive real-world AI testing data. Internal use of AI tools generates feedback that can improve external products, creating a reinforcing cycle of AI capability advancement.

What This Means

Google’s 75% AI code generation milestone signals that AI-assisted development has moved beyond experimentation to core business operations at major tech companies. This level of adoption suggests other software companies will face pressure to match Google’s AI productivity gains or risk competitive disadvantage.

The Deep Research agents target enterprise customers seeking to automate complex research workflows. Success in this market could position Google’s AI infrastructure as essential for knowledge-intensive industries, creating new recurring revenue opportunities.

Isomorphic Labs’ approaching clinical trials represent a crucial test for AI’s pharmaceutical potential. Positive results could validate AI drug discovery and accelerate adoption across the biotech industry. Failures, however, might temper expectations for AI’s near-term impact on drug development timelines and costs.

The convergence of internal AI adoption, enterprise AI products, and specialized applications like drug discovery demonstrates how AI capabilities are becoming foundational infrastructure rather than standalone features.

FAQ

How does 75% AI-generated code compare to other tech companies?
Google appears to lead in AI code generation adoption. While companies like Microsoft and Meta use AI coding tools, none have publicly disclosed comparable adoption rates. The 75% figure likely includes various forms of AI assistance, from code completion to automated testing.

When will Deep Research agents be available in the consumer Gemini app?
Google has not announced plans to bring Deep Research capabilities to the consumer Gemini app. The agents currently require API access and appear designed primarily for enterprise use cases that justify the additional complexity.

What makes AI-designed drugs different from traditional pharmaceutical development?
AI-designed drugs use computational models to predict molecular behavior and identify promising compounds before laboratory testing. This could potentially reduce the time and cost of early-stage drug discovery, though clinical trials still require the same safety and efficacy validation as traditional drugs.