Google's Ex-CEO Just Called this New Way Engineers Work 'Mind Boggling'
Ex-Google CEO Eric Schmidt described the current transformation of engineering work as 'mind-boggling,' highlighting how engineers now delegate coding tasks to AI agents that work overnight while they sleep. This shift is reshaping how engineering teams operate, with companies like Notion and Intercom reporting significant adoption of AI-generated code.
Summary
Eric Schmidt spoke at a tech conference about the dramatic transformation in how engineers work, describing a new workflow where engineers write specifications and evaluation functions, then let AI agents work through the night while they sleep. He believes this will create a bifurcated market with very large companies and very small ones, as the middle ground requiring many average engineers compresses. Real-world examples include Notion's co-founder not writing code for 9 months and Intercom reporting 90% AI-authored pull requests. Companies are productizing this trend, with Perplexity launching 'Computer' - a dedicated Mac Mini that works 24/7 as a digital proxy. However, Harvard Business Review research warns of 'AI brain fry' - cognitive fatigue from managing too many AI tools, with productivity gains plateauing after using three tools. In the competitive landscape, Claude, Replit, and Google are battling for workplace AI supremacy with new integrations and features. Consumer AI data shows ChatGPT maintaining dominance with 900 million weekly users, while Claude and Gemini are seeing rapid paid subscriber growth, suggesting a mobile OS-style duopoly rather than search engine monopolization.
Key Insights
- Schmidt argues that the best engineers with strong mathematical reasoning will become more valuable because someone needs to understand and control AI systems at scale, while below that top tier the calculus changes
- Schmidt describes a working engineer's new workflow where they write specs and evaluation functions, turn on agents at 7pm, and wake up to see what's been built overnight as exactly how startup engineers now operate
- Harvard research reveals that AI brain fry is distinct from burnout, specifically referring to cognitive fatigue from monitoring AI tools beyond natural capacity, with productivity gains plateauing after using three or more AI tools
- Perplexity claims their Computer system saved internal teams $1.6 million in labor costs and performed the equivalent of 3.25 years of work in just 4 weeks across over 16,000 queries
- A16Z researchers argue that AI usage is increasingly undercounted because measuring by web traffic misses embedded AI usage, such as developers spending 8 hours in Claude code
Topics
Full transcript available for MurmurCast members
Sign Up to Access