In an era of expanding trade restrictions and geopolitical disruption, timely commodity and trade indicators are essential for understanding emerging supply-chain exposure.

SCAR is an operational supply-chain early-warning radar that combines deterministic KPI alerting with agentic RAG. A serverless AWS pipeline ingests and normalizes heterogeneous trade, energy, agriculture, price, and transport data from analyst-approved sources. These sources differ in schema, entity granularity, publication frequency, time lag, revision behavior, and geographic/product coverage from near-real-time market or news signals to monthly, quarterly, and annual official statistics.

The deterministic engine aligns these inputs into supply-chain health indicators that surface anomalous conditions. An Amazon Bedrock–based agent router then directs questions to analyst-approved structured tables in Amazon Athena or curated intelligence sources, returning responses with traceable evidence for analyst review. This case study shows how governed retrieval, temporal data normalization, and agentic tool routing can turn fragmented multi-source data into trustworthy, source-grounded supply-chain intelligence without relying on unconstrained model knowledge.