Modern business decision-making is often slowed not by the lack of information, but by the time required to locate, verify, and analyse relevant data from large and complex documents. Financial reports, annual filings, and business records contain valuable insights, yet extracting reliable information from them remains a challenging and time-consuming task. This paper presents Strat-O-Graph, a multi-agent decision-support system designed to simplify financial document analysis through grounded information retrieval and transparent computation. The proposed system combines a React-based user interface, a Spring Boot backend, and a Python-powered intelligence layer incorporating Retrieval-Augmented Generation (RAG), Pydantic-validated AI agents, and a secure computation engine. By integrating document retrieval, automated analysis, and source-linked visualisation, the framework enables users to trace every generated insight back to its original evidence, improving both transparency and trust. The system was evaluated using real-world 10-K financial filings along with the FinQA and TAT-QA benchmark datasets. Experimental results demonstrate significant improvements in information extraction and numerical reasoning accuracy, while practical deployment studies indicate a noticeable reduction in analysis time for end users. The paper discusses the overall architecture, implementation details, service integration, deployment process, and performance evaluation of the proposed framework. The results suggest that Strat-O-Graph can serve as an effective and auditable solution for AI-assisted financial document intelligence and business decision support.

Financial Document Analysis,Retrieval Augmented Generation,Multi-Agent Systems,Auditable AI,Deterministic Computation,Decision Intelligence