"Almost a year into my agentic journey, I’m now quite confident that I don’t need to know every line of code. I get that confidence by knowing that the system behaves as intended, remains maintainable, and lives within the established boundaries."
This is insanity, and at the same time, the strongest and best argument to fire all these useless humans.
"Almost a year into my agentic journey, I’m now quite confident that I don’t need to know every line of code. I get that confidence by knowing that the system behaves as intended, remains maintainable, and lives within the established boundaries."
This is insanity, and at the same time, the strongest and best argument to fire all these useless humans.
> This is insanity, and at the same time, the strongest and best argument to fire all these useless humans.
Well, there's still the human in the loop, uncertainty and risk drive the reviews. From the article:
"I never read all AI-generated code. But, and this is important, make the code you _do_ read count."
The key to make this work is to establish human/agent abstraction boundaries and enforce what you don't inspect.