@atmoio You can try it right now. Ask for an explanation of something you don't understand. Instead of trying to bring your own original ideas to life. It will be clear that the difference is substance. Not quality.
@atmoio Since writing as a craft is the least constrained, the bar for good writing is essentially "what will the reader take away/remember a year after reading this". Nothing on the page itself.
If the writer can bring that, the LLMs are perfectly capable of excellent writing.
@karenzzhou while waiting for subagents to finish. It triggers itself when they do and continues. Every other model I've used gives the main agent something to do while waiting. It feels so weird but it works. Have not seen a session where the main agent failed to pick up and finish the job.