No. They make context usage more efficient, but they're not providing new capabilities that didn't previously exist in an LLM system that could run commands on a computer.