AI tools like Cursor, GitHub Copilot, Windsurf, and Aider have seriously upgraded how we code locally. They're not just fancy autocomplete - they can generate complex functions, handle refactors, help write tests, and cut down on repetitive work. If you've tried them, you know they're game-changers.
Agentic tools like Claude Code, OpenAI Codex, and Jules from Google are going even further, and can complete tasks and open PRs with minimal human assistance.
But as great as these tools are, they're limited to local development. And let's face it, writing code is only part of our job. We’ve all heard the claim for years: developers spend surprisingly little time coding. But has that changed recently? According to two major studies released this year, the answer remains a clear no.
According to a February 2025 IDC Research report, developers spend only about 16% of their time actually building applications. Most of our time goes into supporting tasks like managing CI/CD pipelines, debugging builds, fixing infrastructure problems, and dealing with operational issues.
Microsoft's internal Time Warp: The Gap Between Developers’ Ideal vs Actual Workweeks in an AI-Driven Era survey from April 2025 supports this too, saying explicitly:
"Developers spend only 11% of their week writing code. The majority of their time is fragmented across debugging, meetings, communication, and operational workflows."
Think about your day: how often are you stuck waiting on slow CI pipelines, chasing down confusing test failures, or working through endless rounds of code reviews? These tasks can quickly feel repetitive and drain productivity.
Given how well AI tools have improved local coding, there's clearly a huge opportunity to bring similar improvements to the rest of our workflows, especially around pull requests, continuous integration, testing, and DevOps tasks.
That's why we're working on an AI DevOps Agent explicitly designed for these pain points. Imagine PRs that automatically include clear explanations for failing tests, proactive alerts about CI issues, or contribution guidelines enforcement.
AI's immediate potential for software engineering goes way beyond directly coding. By bringing intelligent assistance into these other areas, we can reduce manual busywork, boost productivity, and make our daily work a whole lot better.
Improving developer experience isn't just about faster local coding. It's about streamlining the entire workflow, from commit to deployment.
The waitlist for our AI DevOps Agent is now open. We will reach out when we're ready to start your onboarding.
