Ask HN: What is shared across participants within "AI-native" environments?
4 comments
One practice I've heard is reviewing formal work specs like https://specs.md/ or https://github.github.com/spec-kit/ INSTEAD of code. I see value in using a standardized spec throughout a company, like a PR template, but skipping code review is a bridge I'm not ready to cross yet.
Me neither.
I feel like the opportunities of learning "how to define problems and establish standards" are getting replaced with "how to write better instructions".
Less about architectures or principles, more and more about prompting, agents, MCPs.
Less about what and how "the work" should be done, becoming more what and how "the instructions" should be written for AI.
Am I the only one experiencing it?