I have a substantial amount of product data, approximately tens of thousands of characters, which I intend to feed into an AI system for intelligent customer service. While I'm aware that some details will require further exploration, I'm seeking industry insights on suitable solutions. The specific challenges I'm facing are: firstly, resource consumption may become overwhelming if I need to provide the AI with the entire dataset every time a conversation is initialized, and secondly, I'm concerned about the risk of the AI being sidetracked to perform non-customer service related tasks, thus wasting tokens. I'm unsure if there are existing solutions or products that cater specifically to these issues. Most of the resources I've found online seem to be traditional chatbots based on word segmentation, whereas I'm interested in exploring solutions built around large language models. Any guidance or recommendations would be greatly appreciated.