You can fine tune a model with specific stuff today. OpenAI offers that right on their website and big companies are already taking advantage. It doesn't take a whole new LLM, and the cost is a pittance in comparison.