We have reached a pivotal time in AI development, with many companies trying to grab mindshare over the right direction for LLMs in coding. Last year, I remained largely unmoved about the outcome until I started seeing code completion finish off a few of my methods, based solely on the cues of standard nomenclature and past patterns.
But today I see that the command line is the best place for the majority of LLM interaction for most developers. I’ve seen multiple offerings that prove the point — from Anthropic’s Claude Code, Open AI’s…