How software engineering is changing: a tale of two eras
AI Summary
An engineer with experience across Bloomberg, PostHog, and Tano shares statistical insights and qualitative reflections on how AI tools like Claude Code Desktop transform software development. By leveraging AI agents, the author experienced massive productivity gains, writing significantly more code and PRs per month while eliminating traditional bottlenecks like communication overhead and human code reviews. However, this shift introduces new engineering constraints, including context switching, diminished shared team context, siloed ownership, and a higher bus factor.
Key Takeaways
- AI-driven workflows allow a single engineer to handle end-to-end full-stack development, compressing months of traditional work into weeks.
- Traditional bottlenecks like manual code reviews and heavy communication overhead disappear, replaced by new constraints like context silos and CI pipeline limits.
- Mitigating a low bus factor in the AI era requires explicitly documenting context in repository configuration files like AGENTS.md.
Why it matters
Understanding how AI alters developer productivity, team processes, and engineering culture helps teams adapt to shifting bottlenecks—such as CI pipelines and source control limits—while managing burnout and maintaining code quality.
Conversation 0
Be the first to share your perspective.