Rene van Osnabrugge discusses how AI is often misapplied to speed up coding, which is only a minor part of development. He suggests AI’s real power lies in reducing bottlenecks in the broader DevOps workflow.

Don’t Let AI Optimize the Wrong 30%

Author: Rene van Osnabrugge

While the current conversation around AI in software development centers mainly on helping developers write code faster, Rene van Osnabrugge argues that this approach is too narrow. Coding might only represent 30% of a developer’s week; the remaining 70% involves meetings, clarifying requirements, documentation, fixing CI pipelines, testing, and handling security alerts. When companies deploy AI for code writing—autocompletion, boilerplate generation, or refactoring—they ignore the larger opportunity: alleviating the routine, less enjoyable work.

Beyond Coding Efficiency

  • Focusing AI tools on programming speed can increase output but adds to tasks like reviewing, testing, and documentation, potentially creating more bottlenecks.
  • Genuine productivity gains come from targeting AI at broader workflow inefficiencies.

Where AI Can Truly Transform Development

  • Requirements Gathering: AI can turn messy meeting transcripts into actionable user stories with clear acceptance criteria.
  • Documentation: AI can auto-generate diagrams and keep technical docs aligned with current code.
  • Testing: Using AI for exploratory testing not only uncovers functional issues but can also produce narratives and screenshots, easily converting them into automated tests for CI.
  • CI/CD & Security: AI can read pipeline logs, highlight likely root causes, suggest patches, and even prepare fixes for alerts such as insecure logging or bad inputs.

Reducing Friction, Not Just Accelerating Output

AI’s most meaningful role is as a friction reducer—clarifying, recommending, summarizing, and automating the mundane. This shift gives developers more time to focus on creative and impactful tasks, restoring satisfaction and flow to their work.

The Real Opportunity for Teams

  • Rather than simply speeding up coding, AI should target the larger chunk of the developer’s job that involves bureaucracy and overhead.
  • Doing so not only avoids piling up unreviewed work but measurably increases team satisfaction and output quality.

“The future of software development is not just faster coders. It is fewer bottlenecks, less friction, and more room for the work that actually inspires people to build.” — Rene van Osnabrugge

This post appeared first on “René van Osnabrugge’s Blog”. Read the entire article here