I can see running a local less resource intensive LLM trained to strip out marketing spiel from the text delivered by the more powerful cloud service LLM being

I can see running a local less resource intensive LLM trained to strip out marketing spiel from the text delivered by the more powerful cloud service LLM being a possibility.