Spending transaction monitor: Agentic AI for intelligent financial alerts

The spending transaction monitor is a Red Hat AI quickstart that uses agentic AI to replace rigid rule-based financial alerting with natural language-driven monitoring. Users describe alert conditions in plain English (e.g., 'notify me if dining exceeds my 30-day average by 40%'), and a LangGraph-orchestrated multi-agent pipeline handles classification, SQL generation, validation, and duplicate detection. The system also recommends new alert rules by learning from historical spending patterns. Built on OpenShift with a stack including FastAPI, PostgreSQL, pgvector, React, Keycloak, and Kubeflow pipelines, it provides a deployable reference architecture for production-grade agentic AI applications in fintech.