Google CEO Says 75% of Company Code AI-Generated as Deep Research Agents Launch

Google CEO Says 75% of Company Code AI-Generated as Deep Research Agents Launch

Google CEO Sundar Pichai revealed Monday that artificial intelligence now generates 75% of the company’s code, coinciding with the launch of two new autonomous research agents that can analyze both public web data and private enterprise information through a single API call.

The disclosure came as Google unveiled Deep Research and Deep Research Max, built on the company’s Gemini 3.1 Pro model. According to VentureBeat, the agents can “fuse open web data with proprietary enterprise information through a single API call, produce native charts and infographics inside research reports, and connect to arbitrary third-party data sources through the Model Context Protocol (MCP).”

The 75% figure represents a significant milestone in enterprise AI adoption, demonstrating how Google has integrated AI code generation across its engineering operations. Multiple Indian publications, including The Times of India and The Indian Express, reported on Pichai’s statement about the company’s internal AI usage.

Deep Research Agents Target Enterprise Workflows

Google positioned the new research agents as infrastructure for “enterprise research workflows in finance, life sciences, and market intelligence — industries where the stakes of getting information wrong are extraordinarily high,” according to VentureBeat.

The agents can autonomously conduct multi-source research that traditionally required hours or days of human analyst time. Deep Research focuses on speed and efficiency, while Deep Research Max offers enhanced capabilities for complex research tasks.

“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. “Use Deep Research when you want speed and efficiency.”

The Model Context Protocol integration allows the agents to connect with third-party data sources, expanding their research capabilities beyond Google’s own data ecosystem. This represents Google’s attempt to position its AI infrastructure as the backbone for enterprise research operations.

Employee Resistance to Military AI Contracts

Google faces internal opposition to potential military AI applications, with over 600 employees signing a letter demanding CEO Sundar Pichai block Pentagon use of Google AI models for classified purposes. According to The Washington Post, many signers work in Google’s DeepMind AI lab, including more than 20 principals, directors, and vice presidents.

The letter states: “The only way to guarantee that Google does not become associated with such harms is to reject any classified workloads. Otherwise, such uses may occur without our knowledge or the power to stop them,” The Verge reported.

This internal resistance echoes Google’s 2018 decision to withdraw from Project Maven, a Pentagon AI program, following employee protests. The current letter suggests ongoing tensions between Google’s AI capabilities and employee concerns about military applications.

Anthropic is currently in a legal battle with the Pentagon over similar issues, according to the sources, though details of that dispute were not fully elaborated.

API-Only Launch Draws User Criticism

The Deep Research agents are available exclusively through Google’s API, not in the consumer Gemini app, drawing criticism from users expecting broader access. Several users on X noted the limitation, with TestingCatalog News observing “Not on Gemini app,” while others suggested Google was “punishing Gemini App Pro subscribers.”

The API-first approach aligns with Google’s enterprise focus for these research capabilities. Logan Kilpatrick, who leads developer relations at Google, noted that the “API has gained a ton of traction over the last 3 months” and expressed excitement for users to “test out the new agents and all the improvements.”

Some users also raised concerns about Google’s presentation of benchmark results, with critics arguing the company’s charts could be “misleading” in how they display performance comparisons.

Google’s AI Code Generation at Scale

The 75% AI-generated code figure positions Google as one of the most AI-integrated technology companies in terms of internal operations. This level of AI adoption in software development suggests significant productivity gains and potential cost reductions in engineering operations.

Google has not disclosed specific metrics about productivity improvements or cost savings from AI code generation. The company also has not detailed which types of code are primarily AI-generated versus human-written, or how quality control processes work for AI-generated code.

The disclosure comes as other major technology companies, including Microsoft and Amazon, have also increased their use of AI coding assistants and automated code generation tools across their engineering teams.

What This Means

Google’s 75% AI code generation rate and the launch of enterprise-focused research agents signal the company’s commitment to AI-first operations both internally and for enterprise customers. The combination demonstrates how Google is using its own AI tools while building products for similar enterprise adoption.

The employee resistance to military AI contracts highlights ongoing tensions between AI capability development and ethical concerns about government applications. This internal opposition could influence Google’s decisions about future government contracts and AI deployment policies.

The API-only launch strategy for Deep Research agents suggests Google is prioritizing enterprise and developer customers over consumer applications for advanced AI research capabilities. This approach may reflect higher revenue potential from enterprise customers compared to consumer users.

FAQ

What percentage of Google’s code is now AI-generated?
Google CEO Sundar Pichai revealed that 75% of the company’s code is now generated by artificial intelligence, marking a significant milestone in enterprise AI adoption for software development.

What are Google’s new Deep Research agents?
Deep Research and Deep Research Max are autonomous research agents that can analyze both public web data and private enterprise information through a single API call. They can produce native charts and infographics and connect to third-party data sources through the Model Context Protocol.

Why are Google employees opposing military AI contracts?
Over 600 Google employees, including many from DeepMind, signed a letter demanding the company reject Pentagon use of its AI models for classified purposes. They argue this is the only way to prevent potential harmful applications without their knowledge or ability to intervene.