Overview
- Business Insider reports that some Google DeepMind engineers can use Anthropic’s Claude for coding, while most other teams are limited to Google’s own Gemini tools, creating friction over perceived performance gaps.
- Engineers told the outlet that Claude works better for coding than internal options in some cases, and they said restricted access has led to frustration on teams that cannot use it.
- Google is pushing wider AI use across engineering, with some staff given specific AI goals tied to performance reviews that include using AI to write code or building tools that speed up their workflows.
- Former Googler Steve Yegge alleges a two-tier system and says anonymous employees told him DeepMind engineers rely on Claude daily, a claim that remains unverified and includes his account of internal pushback when equal access was discussed.
- Google leaders have pushed back in public, with DeepMind CEO Demis Hassabis calling Yegge’s post false and Google Cloud director Addy Osmani saying more than 40,000 company software engineers use agentic coding weekly, a metric Yegge disputes as shallow use.