A surprising revelation from AI leader Anthropic has sparked widespread discussion in the tech world. The company's advanced artificial intelligence, known as Claude Code, is now responsible for generating almost all of its internal software code. This significant development raises an obvious question: why does Anthropic continue to advertise over 100 open engineering positions?
The Evolving Role of Human Engineers
Anthropic executives have stepped forward to clarify this apparent paradox, emphasizing that human engineers remain absolutely vital, even as AI takes over repetitive coding tasks. The shift redefines what it means to be a software engineer in an AI-driven environment.
Engineers Are Moving "Up the Stack"
Boris Cherny, who heads the Claude Code initiative, addressed the issue directly. He explained that human engineers are now focusing on higher-level responsibilities:
- Prompting the AI: Guiding Claude with precise instructions to generate the desired code.
- Customer Interaction: Understanding user needs and translating them into product requirements.
- Team Coordination: Working with different departments to ensure seamless integration.
- Strategic Decisions: Determining what new features or products to develop next.
"Engineering is undergoing a transformation, and skilled engineers are more essential than ever," Cherny stated.
AI Building AI: The Anthropic Approach
At a recent Cisco AI Summit, Anthropic's Chief Product Officer, Mike Krieger, confirmed that "Claude products and Claude code are being entirely written by Claude." This statement aligns with earlier predictions from CEO Dario Amodei, who foresaw AI handling 90% of coding โ a figure that Anthropic has now surpassed internally.
Cherny further elaborated in an interview, revealing his own experience: he hasn't manually written a single line of code in over two months. Instead, he manages between 22 and 27 pull requests daily, all of which contain 100% Claude-generated code. This demonstrates a profound change in daily engineering workflows.
What This Means for Software Development
The practical implications are clear: human engineers at Anthropic are now freed from the "grunt work" of coding. Tasks such as drafting new features, refactoring existing code, and writing documentation are now handled by Claude. This allows human talent to concentrate on:
- System Design: Architecting the overall structure of software systems.
- Architectural Decisions: Making high-level choices about how software components interact.
- Customer Conversations: Gathering feedback and shaping product direction.
- Innovation: Deciding the strategic roadmap for future development.
It's important to note that every piece of code produced by AI still undergoes rigorous human review and validation before it is deployed.
Industry-Wide Impact and Future Outlook
While Anthropic leads the charge with near 100% AI-generated internal code, other major tech companies are also seeing significant contributions from AI:
| Company/Context | Approximate AI Code Generation |
|---|---|
| Microsoft | ~30% |
| Salesforce | ~30% |
| GitHub (Python functions, US) | ~29% (according to a Science journal study) |
Cherny believes that this steep curve of AI adoption in coding, currently observed in "frontier labs" like Anthropic, will soon become the norm across the entire software development industry. This trend signals a fundamental evolution in how software is created, highlighting a future where human ingenuity and artificial intelligence collaborate to build the next generation of technology.
