Amazon Quick and NVIDIA NeMo Agent Toolkit for Specialized Agent Workflows

The post describes a solution that combines Amazon Quick and NVIDIA NeMo Agent Toolkit to create specialized agent workflows for business users, focusing on a supply-chain risk use case. Amazon Quick serves as the front-end for business users, providing a conversational workspace to access structured data, enterprise knowledge from sources like Amazon S3, Google Drive, SharePoint, Confluence, and internal web content, and over 100 pre-built action connectors to third-party systems like Outlook, Slack, Jira, and Asana. Users can also invoke agentic workflows via Model Context Protocol (MCP). NVIDIA NeMo Agent Toolkit is an open source, framework-agnostic library for building, evaluating, profiling, and optimizing agentic workflows, compatible with LangChain, LlamaIndex, CrewAI, Microsoft Semantic Kernel, Google ADK, and custom Python agents. In the solution, supply-chain analysts use an Amazon Quick chat agent to diagnose issues, which triggers a NeMo Agent Toolkit workflow that investigates disruptions, calls supply-chain tools (e.g., purchase order risk, inventory exposure, customer impact, policy lookup, logistics options, mitigation recommendation), validates the recommendation, and returns a ranked mitigation plan with evidence, trace, latency, and evaluator metadata. The architecture includes Amazon Quick for UI and knowledge access, Amazon Bedrock AgentCore as MCP gateway and runtime, and NeMo Agent Toolkit for the backend workflow. Deployment uses AWS CloudFormation, AWS CodeBuild, and Amazon ECR. Users can set up a dashboard using sample data and create knowledge sources for unstructured context.