Case Studies Blog Careers Technologies
Chatbot for Domain based Complex Documents

A Generative AI based to solution for creating a chatbot which can understand complex domain based documents and answer questions based on it.

Problem Statement

The client gets hundreds of requests on a daily basis which require his employees to go through complex documents - Credit Operating Document with 100s of rules to give a very critical decision for his business.

A slight change in understanding of any one of the rules can mean a big difference in terms of revenue for the client. Hence, the client was looking to automate this process so that this happens in minimum time with maximum accuracy freeing up his employees for other complex tasks.


We finetuned the Llama 2 7B model on the Credit Operating Document such that it is able to understand the context of various questions asked to it as map it to the document and give the relevant answers accurately. It was able to understand the data - tabular, pointer based and in any other format


  • Development: Python
  • LLM: Llama 7B

Value Delivered

Our model is accurately answer any question based on any data present in the document - be it structured or unstructured