Loading video player...
This project tackles the problem of slow, single-perspective scientific literature review by automating the research synthesis process. It addresses the challenge of integrating knowledge from vast and varied sources (PDFs, arXiv, Wikipedia) into a coherent analytical output. During the hackathon, we built a Multi-Agent Orchestration API using FastAPI, LangChain, and OpenAI (GPT-4o-mini). This system performs Retrieval Augmented Generation (RAG) and simulates a scientific debate between four specialized AI agents (e.g., Mathematical, Sociological, Physical Analysts). Academic researchers, corporate R&D teams, and specialized analysts benefit from the solution by getting accelerated insights. What works today is the full end-to-end pipeline: knowledge ingestion, vector store creation, multi-agent discussion, and the final generation of a structured PDF scientific report complete with a synthesized hypothesis, conclusion, and a dynamic knowledge graph visualization. This results in multi-disciplinary analytical reports generated in minutes, showcasing a highly effective and innovative use of AI collaboration.