FinLedger AI: A Blockchain-Secured Organizational Finance Monitor with LLM-Powered Analysis

Abstract Most organizational finance systems share three persistent weaknesses: historical records can be altered without leaving a trace, users are granted broader access than their roles require, and the systems are reactive - they store and calculate but do not warn anyone when something looks wrong. This paper presents FinLedger AI, a web-based financial ledger that addresses all three problems in a single system. Built on the MERN stack (MongoDB, Express.js, React, Node.js), FinLedger AI seals every transaction into a custom blockchain in which each block is hashed with SHA-256 and linked to its predecessor, so any after-the-fact modification breaks the chain and is flagged immediately during verification. Access is separated by role: administrators see organization-wide data, while staff see only their own records. On top of this integrity layer, a large language model (Google Gemini) continuously analyses the ledger to produce budget alerts, spending forecasts, and plain-language financial summaries. Tamper-detection tests confirmed that altering even a single stored value breaks the hash chain and surfaces an alert. The system demonstrates that tamper-evident record keeping and automated financial intelligence can be combined in one practical application built entirely on an open-source stack.