Post

As a programmer today, it's basically a given I will use LLMs both (1) to write 'new' code and (2) to make the programs I write make further requests to LLM service (API) providers. Starting to seem like I have a moral prerogative to choose the less evil provider, wherever possible

I respect that. Some things should absolutely not be vibe-coded. I believe skilled programmers (can) make much better "prompt engineers" similar to how good programmers (can) make better engineering managers: It helps to be able to tell when your "team" (LLM) isn't producing to your standards

Agree 100% I’ve been altering my stack to mimic HumanLayer’s, because I believe that the (very near) future will be working with LLMs to write the code, while I write the specs and guardrails and keep everything aligned. github.com/humanlayer/h...

github.comGitHub - humanlayer/humanlayer: The best way to get AI coding agents to solve hard problems in complex codebases.The best way to get AI coding agents to solve hard problems in complex codebases. - humanlayer/humanlayer

0 reposts2 likes

No replies yet.