an example of this is bug fixing you typically would think of

Press Space to continue

an example of this is bug fixing you typically would think of the most obvious cause and check. then repeat until you find it the agent can come up with all potential causes at once, instrument the code and execute it once to figure out which one caused it

dax@thdxr4h ago

one side effect of treating LLMs like humans is people keep using them ineffectively you can't break a task up into 5 pieces and work on all of them in parallel and non-linearly but the agent can. you don't even have to tell it to, if it has the right environment it will do it