AI is writing 75% of Google’s code now — engineers just approve it, reveals Sundar Pichai

According to CEO Sundar Pichai, around 75 per cent of all new code at Google is now generated by AI systems and subsequently reviewed by human engineers.

This marks a sharp increase from about 50 per cent reported last year.

Pichai shared the update in a post on The Keyword, highlighting how AI has become central to Google’s internal engineering workflows.

Engineers shift from coding to reviewing

The company said the role of engineers is evolving, with a growing focus on validating and refining AI-generated code rather than writing it from scratch.

This shift reflects what Google calls “agentic workflows,” where AI tools handle large portions of technical tasks while engineers supervise and coordinate outputs.

AI agents speed up complex development tasks

Sundar Pichai also pointed to internal experiments where AI agents worked alongside engineers to complete complex code migration projects up to six times faster than human-only teams had managed previously.

He described engineers as now “orchestrating autonomous digital task forces,” using multiple AI agents to execute and streamline development work.

Gemini models drive internal AI push

Google said its AI transformation is being powered largely by its Gemini models, with increasing adoption across teams.

The company also reported strong growth in its Gemini Enterprise platform, which saw a 40 per cent rise in paid monthly active users quarter-over-quarter.

The shift is also raising new challenges, particularly around managing large-scale deployment of AI agents across teams.

Tech industry race to expand AI coding

Google’s move comes amid broader industry-wide adoption of AI in software development. Microsoft has also reported significant AI involvement in coding, with executives suggesting that a growing share of code in the future could be machine-generated.

The trend signals a wider shift across Big Tech, where AI is increasingly moving from a support tool to a core part of software engineering.