Streamlining Financial Analysis for a Global Bank using ORIAN-GenAI Accelerator

Challenges:

The client is a global leader in the banking and financial services domain. They were investing significant time in mining, analyzing, and comparing diverse document types of competitor data for financial analysis. They wanted to leverage GenAI for automated data extraction from complex financial statements but were saddled by limited cloud connectivity in their infrastructure setup. They grappled with:

• Difficulty in managing large data volumes and documents from diverse sources.
• Complicated manual analysis resulting in errors and delays.
• Restricted access to cloud connectivity and open-source packages.

Solution:

Xoriant enabled the client with a Knowledge Management Platform, harnessing a Retrieval Augmented Generation (RAG) pipeline on its GenAI accelerator, ORIAN. This solution analyzes competitor documents, leveraging Large Language Models (LLMs) for precise insights.

• Implemented Retrieval Augmented Generation (RAG) Pipeline with ORIAN, Xoriant's GenAI accelerator.
• Leveraged Open-Source Technology on virtual machines hosted on private cloud.
• Extracted Metadata with specialized PDF table extraction packages.
• Employed Large Language Models (LLMs) for extracting table commentary, enhancing inference capabilities.
• Simplified Document Retrieval by breaking down complex queries into a list of sub-questions.
• Streamlined Information extraction by concatenation of results from all sub-answers.

Business Value:

• Achieved 80% user satisfaction in query responses.
• Facilitated informed decision making with improved accessibility to critical information.
• Enhanced competitive analysis capabilities.

Technology Stack:

Generative AI | LLM | Python | React JS | MongoDB | Flask | Nginx | Ubuntu | Llama-2 | Mixtral 8X7B | Chroma DB | BGE Embeddings | LangChain | Chain-of-Thought | Streamlit | Ragas