Guidance For Deploying AI Agents To Device Fleets Using AWS IoT Greengrass — DevOps skill for Claude Code

Guidance for Deploying AI Agents to Device Fleets Using AWS IoT Greengrass

Table of Contents

1. Overview
   • Cost
2. Prerequisites
   • Operating System
3. Deployment Steps
4. Deployment Validation
5. Running the Guidance
6. Next Steps
7. Cleanup
8. FAQ, known issues, additional considerations, and limitations
9. Notices

Overview

This Guidance demonstrates how to deploy AI agents to edge device fleets using AWS IoT Greengrass. It addresses the challenge of running intelligent, multi-agent systems at the edge where devices need to process queries locally using small language models (SLMs) while maintaining connectivity to AWS cloud services.

The solution uses StrandsAgents, a lightweight framework that enables multi-agent systems with local inference capabilities. An orchestrator agent intelligently routes queries to specialized sub-agents that can retrieve real-time data from industrial equipment via OPC-UA protocol and consult technical documentation to provide comprehensive responses. This architecture is ideal for industrial IoT scenarios where low-latency responses, offline operation capability, and data sovereignty are critical requirements.

Architecture


**Architecture Flow:**

1. User queries are published to AWS IoT Core MQTT topics
2. AWS IoT Greengrass component receives the query on the edge device
3. The orchestrator agent analyzes the query and determines which specialized agents to invoke
4. Sub-agents execute their tasks:
   • OPC-UA agent retrieves real-time equipment data from local industrial devices
   • Documentation agent searches technical manuals for relevant information
5. Local Ollama inference engine processes agent requests usin