Build an AI Customer Support Agent on AWS

🚀 What You'll Build

You will build an AI customer support agent that reads your product documentation (PDFs/text), answers customer questions using that knowledge, remembers each customer across sessions, and can create support tickets or escalate to a human — all running on Amazon Bedrock AgentCore. You build and deploy it with the AgentCore CLI (@aws/agentcore) — one command-line tool that generates the agent, connects its tools and Knowledge Base through an AgentCore Gateway, gives it managed memory, and ships it to the cloud. No Docker, and no cloud permissions to set up by hand. A small cross-platform bootstrap script supplies the supporting pieces (sample docs, the tool Lambdas, and helper scripts) identically on Windows, macOS, or Linux.

✅ Prerequisites

• An AWS account with Amazon Bedrock available (us-east-1 recommended)
• Familiarity with the AWS Console and AWS CLI
• Basic understanding of what an LLM/AI model is — you do not need ML expertise
• Python 3.10+ installed on your laptop (3.12+ recommended; Windows, macOS, or Linux — all supported)
• Node.js 20+ installed (22 LTS or newer recommended) — the AgentCore CLI (@aws/agentcore) ships as an npm package
• AWS CLI v2 configured with a profile that has AdministratorAccess
• Git installed (optional — there is a no-Git fallback for the auxiliary scaffold)