OpenAI keeps making agents better at doing work, but hasn't figured out how humans and agents should WORK TOGETHER.
The missing thing is collaboration model: who owns what, when to interrupt, when to act autonomously, and when to shut up / deliver. #dots#codex
How can a company fail once, yet have tons of explanations for why?
Bad product kills conversion. Low conversion drives up CAC. High CAC burns cash. No cash means no product improvements.
All connected. Start anywhere in the loop, and you can trace a complete chain of failure.
“42” is the perfect answer. The problem? The question was so damn underspecified that nobody could make sense of the answer, do anything useful with it, or even verify it was right.
A perfectly good answer to a completely busted spec.
I swear the guy had a time machine.
Spent the weekend rereading The Hitchhiker’s Guide to the Galaxy. Reading it now with all this AI agent stuff going on, it feels like Douglas Adams was writing about the exact mess we're dealing with today.
Spent the weekend rereading The Hitchhiker’s Guide to the Galaxy. Reading it now with all this AI agent stuff going on, it feels like Douglas Adams was writing about the exact mess we're dealing with today.