Agentic AI Foundations

In this course, you’ll explore the core principles and strategies for designing Agentic AI systems using AWS services. You’ll learn how Agentic AI differs from traditional conversational systems, and how to use tools like Amazon Q, Kiro, Amazon Bedrock Agents, and Amazon Bedrock AgentCore to build autonomous, goal-driven solutions that solve real-world problems.

Objectifs de la formation

A l’issue de cette formation AWS, vous serez capable de :

• Summarize the evolution of agentic AI and define what makes something agentic
• Identify the core components of agentic systems
• Distinguish between workflow, autonomous, and hybrid agents
• Identify basic implementation patterns for agentic AI
• Compare AWS service options for agentic AI
• Describe the capabilities and use cases of Amazon Quick Suite, Amazon Q Developer, and Kiro
• Explain Strands Agents framework and its applications
  Build and customize a basic AI agent using Strands Agents
• Develop a simple task-specific agent for real-world application using Strands Agents SDK
• Explain Amazon Bedrock AgentCore
• Describe observability and interoperability patterns for production agentic AI systems

Public

Cette formation s’adresse aux analystes business, support informatique, professionnels du marketing, gestionnaires de produits ou de projets, chefs d’entreprise ou responsables informatiques, commerciaux IT

Prérequis

We recommend that attendees of this course have the following experience:
• Generative AI Essentials or equivalent work experience
• Foundational AWS knowledge and software development experience

Programme de la formation

From LLMs to Agents

• Describe the progression from AI and ML through generative AI to agentic AI
• Explain the limitations of traditional large language models
• Identify the fallacy of composition in LLM applications
• Define what AI agents are
• Describe the key innovations enabling agentic AI

Exploration des cas d’usages de l’IA Générative

• Define what makes an AI system agentic
• Identify the core components of agentic systems
• Differentiate between workflow, autonomous, hybrid, and multi-agent systems
• Describe use cases for each agent type
• Analyze factors driving agentic AI adoption

Understanding Agentic AI Workflows

• Explain serialization (prompt chaining) in agentic workflows
• Describe parallelization patterns for agent tasks
• Identify routing mechanisms for task distribution
• Explain orchestration approaches for complex workflows
• Describe Amazon Bedrock Flows as a managed workflow solution

AWS Agentic Development and Productivity Tools

• Explain Amazon Quick Suite features and applications
• Describe Amazon Q Developer capabilities and use cases
• Explain the Kiro spec-driven development methodology and its agentic capabilities
• Compare Amazon Q Developer and Kiro for different development scenarios
• Analyze when to choose different AWS agentic solutions

Implementing Agentic AI Frameworks with Amazon Bedrock AgentCore

• Compare and contrast open source agentic frameworks
• Describe how to define the components of an agent using Strands Agents
• Describe the challenge of production and the prototype-to-production gap
• Explain Amazon Bedrock AgentCore architecture and its seven core primitives

Building Custom Solutions

• Identify scenarios requiring custom agentic AI approaches
• Apply decision criteria for choosing custom or managed solutions
• Understand CloudWatch generative AI observability for monitoring agentic AI applications
• Summarize observability dashboards for model invocations and agent performance
• Explain the A2A protocol for standardized agent communication.
• Explain MCP concepts
• Summarize MCP with A2A for comprehensive interoperability

Méthodes pédagogiques

Des exercices pratiques et des démonstrations vous permettront de mettre en pratique les notions théoriques présentées.

La dernière version du support de cours, en anglais, vous est transmise par voie dématérialisée. Les cours seront disponibles en ligne pendant 730 jours après leur activation et téléchargeables avec Bookshelf application. Pour y accéder, il est nécessaire de créer un compte eVantage sur evantage.gilmoreglobal.com.

Méthodes d’évaluation des acquis

Afin d’évaluer l’acquisition de vos connaissances et compétences, il vous sera envoyé un formulaire d’auto-évaluation, qui sera à compléter en amont et à l’issue de la formation.
Un certificat de réalisation de fin de formation est remis au stagiaire lui permettant de faire valoir le suivi de la formation.

Accompagnement

Vous avez un projet de migration dans le cloud AWS ? Vous souhaitez être accompagné ? KANOPEE peut vous aider et vous répondre à toutes vos questions sur toutes les différentes étapes. Contactez-nous !