Eternal Wisdom AI: An Intelligent Agentic Retrieval-Augmented Generation System Using NLP and LLMs
Eternal Wisdom AI: An Intelligent Agentic Retrieval-Augmented Generation System Using NLP and LLMs
Prathmesh Abhang, Prajwal Aswar, Rushikesh Ban, Shreyash Deokar, Prof.Shruti Hatwar, Prof.Suvarna.A.Bahir
Department of Computer Science and Engineering, TSSM’s PVPIT, Pune, India
Abstract—This paper presents Eternal Wisdom AI, a novel intelligent agentic system integrating Retrieval-Augmented Generation (RAG), Natural Language Processing (NLP), vector database indexing, and Model Context Protocol (MCP)-based multi-agent orchestration within a FastAPI backend. Existing LLMs suffer from factual hallucination, static knowledge boundaries, and limited autonomous reasoning. Eternal Wisdom AI addresses these through a dual-pipeline architecture: a semantic vector retrieval pipeline and an agentic decision layer that dynamically selects tools, invokes MCP servers, and synthesizes information through chain-of-thought reasoning. Experimental results demonstrate a hallucination rate of 3.2%, task completion accuracy of 87.4%, and faithfulness score of 0.91—substantial improvements over standard RAG baselines.
Index Terms—Large Language Models, Retrieval-Augmented Generation, NLP, Agentic AI, Vector Databases, MCP Protocol, FastAPI.