AI-Powered Finance Chatbot

The client works in the finance industry and we developed an AI-powered chatbot for their website. The goal was to streamline the process of accessing and delivering the exact information based on queries by the end users. By harnessing the power of OpenAI technology, our chatbot drastically reduced the time that was getting consumed earlier for this. To achieve this, we used a Large Language Model (LLM) based AI chatbot. We used OpenAI APIs to seamlessly communicate with the finance data that was provided to us by the client in the form of pdf files. The process involves feeding these files into the AI model, by chunking the texts and then converting it into vector form. OpenAI Embedding is used to feed the information with OpenAI APIs. The method has enabled us to significantly reduce users’ search time by an impressive 90%, sparing them the ordeal of having to manually navigate through numerous website pages to arrive at the information they need.

We trained the chatbot to analyze user queries and extract relevant information from the pdf files provided. The process ensures that the responses offered are those that are extracted solely from the information available in the pdfs only; with nothing coming from the internet. Through seamless integration via API and robust data storage capabilities, our chatbot not only enhances user experience but also provides valuable insights that can be harnessed for analytics downstream.

Client : Confidential

Location: USA

Industry: Financial Services

Python, FastAPI, OpenAI, LangChain, FAISS vector DB, Chroma DB, PyPDF2, python-dotenv, Linux, PostgreSQL, OpenAI Embeddings