The Coming Orchestration Era in Software Engineering

At Anthropic, Boris Cherny—creator of Claude Code—hasn’t written a line of code by hand in months. His team routinely ships dozens of AI-generated pull requests daily, while the company maintains software engineering openings with total compensation packages exceeding $570,000.

This apparent contradiction highlights a critical shift: As AI coding agents become more capable, the value isn’t in who can write code fastest but in who can orchestrate these tools most effectively. NVIDIA CEO Jensen Huang recently echoed this sentiment, suggesting that engineers not leveraging AI coding assistants are like chip designers still using paper and pencil.

From Implementation to Orchestration

The fear that AI will replace software engineers entirely is misplaced—just as ATMs didn’t eliminate bank tellers but rather transformed their role. When ATMs were introduced, teller employment doubled from 1970 to 2006 as banks opened more branches and tellers focused on complex customer interactions.

Similarly, AI coding agents will handle routine implementation tasks (boilerplate code, CRUD endpoints), while human engineers focus on higher-level responsibilities like:

  • Defining what needs to be built
  • Evaluating AI-generated outputs
  • Governing deployments
  • Maintaining increasingly complex systems

This shift explains why Anthropic and others are hiring specialists in orchestration, governance, and prompt engineering—skills that complement rather than compete with AI agents.

The Software Development Advantage

The rapid adoption of AI coding tools in software development highlights a critical advantage: The industry already has the infrastructure needed to support these technologies. Six key components that mature software development practices have solved:

  1. Governance: Clear permissioning frameworks from code creation to deployment
  2. Observability: Comprehensive audit trails for every decision and change
  3. Evaluation: Established processes for reviewing and validating outputs
  4. Security: Robust access controls and vulnerability management
  5. Automation: Integrated CI/CD pipelines for consistent delivery
  6. Version control: Git-based systems that track all changes and allow rollbacks

Enterprises in other functions should view this as a roadmap rather than an endpoint—as they build these capabilities, AI agents will become more valuable across the organization.

The $570K compensation packages aren’t for coders; they’re for orchestrators. As software development leads the way into the age of AI-augmented workforces, other industries should take note: The real value lies not in fearing replacement but in embracing new roles that leverage these powerful tools.