Visão Geral
Este curso apresenta os conceitos, arquiteturas e práticas de orquestração de agentes baseados em LLMs, utilizando frameworks modernos como CrewAI, AutoGPT e LangGraph. O foco está em coordenar múltiplos agentes especializados, garantindo fluxo de trabalho controlado, colaboração eficiente e execução confiável de tarefas complexas.
Conteúdo Programatico
Module 1 – Introduction to Agent Orchestration
- What is agent orchestration
- Why single agents are not enough
- From pipelines to agent crews
- Orchestration use cases
Module 2 – Orchestration Architectures
- Centralized vs decentralized orchestration
- Agent roles and responsibilities
- Task decomposition strategies
- Workflow design patterns
Module 3 – CrewAI Fundamentals
- Crew-based agent concept
- Roles, goals and backstories
- Task assignment and execution
- Collaboration models
Module 4 – AutoGPT and Autonomous Execution
- AutoGPT architecture
- Planning and execution loops
- Memory and tool usage
- Strengths and limitations
Module 5 – LangGraph and State-Based Orchestration
- Graph-based agent workflows
- Nodes, edges and states
- Conditional execution
- Error handling and recovery
Module 6 – Communication, Memory and Context Sharing
- Inter-agent communication
- Shared vs isolated memory
- Context synchronization
- Knowledge propagation
Module 7 – Control, Safety and Observability
- Guardrails and constraints
- Execution limits
- Logging and monitoring
- Debugging agent workflows
Module 8 – Enterprise Use Cases and Best Practices
- Corporate automation scenarios
- Scalability considerations
- Security and compliance
- Future of agent orchestration