CGS ITTraining

Courses / AI Agents with Amazon Bedrock AgentCore

Course · 2 days · on request

AI Agents with
Amazon Bedrock AgentCore

You build an agent that calls your APIs and uses your documents, put critical actions behind an approval gate and run it on AWS.

  • 2 days
  • Python and Java
  • Strands Agents · AgentCore
  • in-house, open or remote
Request course Schedule

Example project

A booking assistant for an existing application

The agent queries free seats through the API of a Java service, answers questions from the course descriptions and creates a booking. The booking is only created after a human approves it at the gate.

Audience

Developers and architects who bring AI agents into existing applications on AWS.

You bring

Programming experience in Java or Python, AWS basics (console, IAM), Git.

Not required

Prior experience with LLMs or agent frameworks.

Schedule: two days, one agent

Day 1

Build the agent

From the first agent to the connected application

  • Bedrock and AgentCoreModels, Converse API, regions and data protection; when a workflow is enough and when you need an agent
  • First agentStrands Agents: model, tools, agent loop; run locally and deploy to AgentCore Runtime
  • ToolsConnect a Java service as MCP tools through AgentCore Gateway
  • Knowledge and memoryYour own documents through a knowledge base, context across sessions with AgentCore Memory

Day 2

Secure and operate

Gates, permissions, measurement, cost

  • GateCritical actions only after approval: human in the loop in the agent workflow
  • SecurityBedrock Guardrails and AgentCore Policy against prompt injection and data leakage; AgentCore Identity for user permissions
  • OperationsObservability and evaluations: trace behaviour, measure quality, detect regressions
  • Cost and integrationTokens, prompt caching, model choice; calling the agent from the Java application

Request the course

We put this course together on request. We agree on content, focus areas and dates with you.

Send e-mail Call WhatsApp

[email protected] · +43 650 944 699 0 (also WhatsApp)