I am considering whether to do RAG or Fine tuning for LLM to answer questions for clients in a small business
The small business is a car rental company that gets
- new clients call up about certain cars in the fleet to discuss renting them for dates X to Y, ask for price,
- existing clients/renters calling to coordinate car pickup, drop-off, finalise contract, insurance etc
The company has data for last 5 years where a customer care team has been fielding these questions, and answering questions based on a set of documents; but also checking the a rota /time table/database about which cars are available, which cars have problems and are out for repair, and consulting documentation about pros/cons of each car
The current customer care also discuss questions one does not know with others in the team in a free flowing slack workspace with around 10 different slack channels (say one for "Repairs", one for "insurance" , one for "security deopsit" etc - there are various slack channels where relevant conversations happen - but these are free flow, they dont always conform to question and answers as these are internal to the team)
I want to explore how to put all the data I have
- Slack archives, perhaps i can dump each channel history for last 5 years as text dump
- Conversation history with clients(where each client emails us and we have threads of email conversation back and forth). These conversations are in a CRM and have rich-data/tags like each conversation is tagged with model of car, dates, price paid, customer satisfaction rating etc
How can I feed all of these to a LLM, and also give it the rota/state of the business for each day - in terms of which cars are avilable are current available and which are booked till what date
and, use it to answer questiosn from new clients?
