"Thinking" models do that, now. They either search for the context (token usage goes brrr...) or ask for more information before going further.
"Thinking" models do that, now. They either search for the context (token usage goes brrr...) or ask for more information before going further.