Why Humans Can't Keep Up With AI-Written Code - Tariq Shaukat, Sonar

By ai.engineer

Categories: AI, Tools

Summary

AI agents are generating 24x more code but developers can't keep up—only 30% increase in actual releases despite 2,450% more lines of code. The solution isn't human oversight but engineering agentic software factories using Toyota production system principles: prioritize quality over volume, fix systems not symptoms, and build deterministic governance into the development pipeline.

Key Takeaways

  1. Code volume explosion creates a shipping bottleneck: 2,450% increase in lines of code, 240% increase in commits, but only 30% increase in releases. This reveals the real constraint isn't code generation but QA, testing, and deployment automation.
  2. Human-in-the-loop doesn't scale with agentic development. Instead, adopt Toyota production system principles: prioritize quality over volume, fix root causes not symptoms, empower automated systems to stop the line, and design for buildability from the start.
  3. Cognitive surrender is real: humans disproportionately agree with AI outputs. Security and governance strategies must use deterministic system design and reliable operating procedures, not manual review gates, to contain insider risk and model misuse.
  4. 34% of developers expect agentic software factories within a year (39% in financial services). This signals an industry shift toward automated CI/CD and governance. Early adopters need engineering-first containment strategies built into their pipelines now.
  5. Security risk doubles as agents scale: same tools enabling rapid development can automate attacks, find vulnerabilities, and cause insider threats. Require deterministic controls and system-level governance architecture, not just access policies.

Related topics

Transcript Excerpt

[music] >> Uh I My name is Tariq Shakir. I am the CEO of Sona, as you just heard. And we're here to talk today about how we have been um helping companies build software factories that actually work for enterprises. And I thought it'd be an interesting way to start to start looking at actual factories cuz we know how to build actual factories. And for those of you who have a little bit of the gray hair and all of this sort of thing, you may uh remember the 1980s. And in the 1980s, you could find cars that were really good cars. But if you look at the picture here and if you remember those times, the cars like didn't quite fit together perfectly. Right? You'd have different pieces that would leave big gaps. It would be really good cars that just didn't kind of work the way they were suppose…

More from ai.engineer