so, i need advice from the community about a system that i am struggling to create.
Let me explain the problem first, so, here it goes.
I am trying to build a mobile app, powered with AI. It is an astrology app. I have all the relevant data as my guru is an expert with 60 years of experience in this field.
I am able to call an astro API to get the user chart details. so, i have the planetary positions, aspects, and other related information available with me for the AI.
the purpose of using an AI is to make predictions. to draft an overall structure of a person based on the above details and the data that I have.
the challenge that I am facing is that, this is not a simple RAG over my dataset. later on, i will come to what all i have already tried. but for now, the challenge is that, for predicting any part of a human life, be it personality, life, career, relationship or anything, there are a set of rules that needs to be followed and checked. for an example, to get the personality, there are around 6 rules to be followed, like the placement of sun, moon. the ascendant, the planets present in the house of personality, the ruler of those houses and where they r placed in the chart, to name a few rules/combinations.
Now, RAG is simple retrieval of completion of the prompt fed into the llm. and as you can see, i need to give reasoning, thinking and processing power to the llm.
I have already tried Llama index, langchain with various retrievals, agents, tools and even with vectors, knowledge graphs, summary tools etc. but no use.
kindly help/guide me solve this issue. any pointers/links will be appreciated.
