Molly Sheets reflects on her career shift from film to tech, emphasizing financial independence and work-life balance, while critiquing AI’s impact on software development and the evolving role of developers. The discussion covers AI’s risks, the decline of traditional coding skills, and the need for systems thinking, interdisciplinary learning, and psychological safety in engineering.

AB Testing

AB Testing - Each episode is a chat between Brent Jensen and Alan Page with an occasional special guest.

• https://www.moderntesting.org/

Episode Details

• Show Notes: https://podcasters.spotify.com/pod/show/abtesting/episodes/Episode-236---Molly-Sheets-e3ojql2
• Published: 2026-09-09T18:26:20Z
• Duration: 00:51:36
• Author: AB Testing

Overview

The podcast centers on reflections about career, work-life balance, and the evolving role of developers in the age of AI. Molly Sheets shares her journey from film to tech, including her experiences at companies like Amazon and Unity, her transition into a sabbatical, and her skepticism about returning to traditional work. She emphasizes the importance of financial independence, the risks of relocating for jobs in high-cost areas, and the erosion of work-life boundaries when personal time is consistently compromised by work demands.

A significant portion of the discussion explores the impact of AI on software development, questioning whether coding skills will remain essential as AI tools advance. The conversation highlights the shift toward systems thinking and the need for developers to manage AI agents rather than write code manually. Concerns are raised about over-reliance on AI-generated code, the decline of computer science degrees, and the value of interdisciplinary learning. Additional topics include modern testing principles, continuous deployment, incident management, and the importance of psychological safety and learning from failure in engineering cultures.

What If

• What if you treated your personal projects like production systems?

  • Move: Implement CI/CD pipelines with automated rollbacks and monitoring for your solo projects using tools like GitHub Actions, Sentry, and health checks.
  • Why Now?: With AI accelerating code generation, the risk of undetected failures increases - automated recovery builds real confidence faster than manual testing.
  • Expected Upside: Reduce debugging time by 50%+ and increase deployment frequency, turning side projects into reliable, scalable offerings that can generate income or attract collaborators.

• What if you replaced traditional testing with confidence engineering in your workflow?

  • Move: Shift focus from writing exhaustive unit tests to designing small blast radius deployments, real-user monitoring, and automated alerts for critical failures.
  • Why Now?: Modern tools (e.g., Vercel, Netlify, Cloudflare) make ephemeral environments and production-like testing accessible to solo developers - mimicking high-performing teams at scale.
  • Expected Upside: Ship features 2 - 3x faster while maintaining stability, freeing up time to build differentiating functionality instead of test scaffolding.

• What if you used AI not to write code, but to stress-test your system thinking?

  • Move: Use AI agents to simulate edge cases, challenge your architecture decisions, and generate failure scenarios before implementation.
  • Why Now?: AI can now act as a low-cost, on-demand systems reviewer - catching gaps in logic or design that lead to costly rewrites later.
  • Expected Upside: Reduce technical debt by 40%+ in early-stage projects and increase the robustness of your software, making solo development more competitive with team-based outputs.

Takeaway

• Prioritize building systems that allow for safe, small-scale failures in production by implementing automated rollbacks and monitoring to quickly detect and resolve issues.
• Focus on reducing the blast radius of deployments through techniques like phased rollouts, IP-based targeting, or cell isolation to minimize user impact.
• Treat every deployment as a potential incident by proactively creating incident channels and involving SRE teams, ensuring real-time response readiness and runbook development.
• Shift from maintaining multiple staging environments to adopting ephemeral, on-demand environments using tools like Docker or Minikube for reliable pre-deployment validation.
• Invest in post-incident learning by analyzing communication (e.g., Slack channels) to extract actionable insights and improve team confidence and psychological safety.