Advanced AutoGen Training in Madrid, Spain

This training is designed for developers and AI practitioners who want to go beyond the basics and build robust, complex, and modular multi-agent applications using the AutoGen framework. Participants will explore advanced use cases, custom agent development, tool integration, agent collaboration strategies, and real-world deployment scenarios.

Advanced Architecture and Design Patterns

Deep Dive into AutoGen Architecture

• Review of AutoGen agents, GroupChat, tools, and memory
• Understanding message flow and callback structure
• Limitations and architecture constraints

Role-Specific Agent Design

• Creating modular agents with unique capabilities
• Example roles: planner, executor, validator, writer, reviewer
• Combining agents for task decomposition and recomposition

GroupChat and Agent Collaboration

• Multi-agent task planning with GroupChat
• Managing turn-based conversations
• Handling conflicting outputs and refining results
• Creating fallback agents and escalation logic

Tool Integration, Memory Use, and Automation

Advanced Tool and API Integration

• Creating and registering custom tools
• Calling external APIs from AutoGen
• Example: real-time data fetching, calendar management, task scheduling
• Tool chaining between agents

Memory Management and Context Handling

• Overview of memory types (in-memory, file-based, vector stores)
• Persisting task history across sessions
• Passing memory between agents for long-term reasoning
• Memory cleanup, context reset, and optimization

Automation Exercises and Deployment

Final Hands-On Exercises