Is AI-Powered Engineering Changing How We Should Build Applications?

Maria

AI coding tools are making application development faster, but speed isn't necessarily the hardest part anymore.

The bigger question is how teams can use AI throughout the engineering lifecycle without compromising architecture, security, testing, or maintainability.

For example, GeekyAnts is exploring an AI-native engineering approach where AI is used across development while human review, testing, security, and observability remain part of the delivery process. Its AntFlow AI also uses specifications, multiple agents, independent verification, and human-controlled delivery.

I'm curious how the open-source community sees this evolving:

Should AI-generated code follow the same review standards as human-written code?
How should projects track AI-assisted contributions?
Could agentic development eventually become part of standard CI/CD workflows?
What safeguards are needed when AI agents can modify repositories or call external tools?

Would you trust an AI agent to contribute directly to a production open-source project if every change still went through the normal review and testing process?