Of course we can code without a LLM, but if we want to, as the sibling comment says, don't run a local LLM on a laptop. I'm offloading the LLM to the LLM vendor, like most of us IMO. The CPU is still busier then it used to be, because of the local harness.