Amazon Quick with NVIDIA NeMo Agent Toolkit for Supply Chain Risk Decision Workflows

This solution uses Amazon Quick as the user entry point, leveraging its knowledge integrations (supporting S3, Google Drive, Confluence, etc.) and over 100 prebuilt action connectors, and invoking the backend NVIDIA NeMo Agent Toolkit workflow via Model Context Protocol (MCP). NeMo Agent Toolkit is an open-source, framework-agnostic library for connecting, evaluating, analyzing, and optimizing agent workflows, supporting LangChain, LlamaIndex, CrewAI, and other major frameworks. The backend workflow runs in an Amazon Bedrock AgentCore Runtime container, where the supply_chain_risk_orchestrator orchestrates six registered functions: purchase order risk, inventory exposure, customer impact, contract policy, logistics options, and mitigation recommendations. Functions are defined by YAML config files, each can be tested, reused, and versioned. The workflow outputs a ranked mitigation plan with evidence, execution trace, latency, and evaluation metadata. For supply chain ops leaders, analysts, or startup teams, this architecture scales operations without adding planning headcount. For developers and engineering leaders, NeMo Agent Toolkit provides telemetry, latency, and evaluation results to fine-tune workflows and tools pre-production.