A2A Custom Agents Workshop

In this workshop, you’ll learn to build custom AI agents using xpander.ai that collaborate through Agent-to-Agent (A2A) protocols. Using coding agents as a practical example, you’ll create agents that work together to handle complex multi-step tasks through intelligent coordination and communication.

When finished, you’ll understand how to create custom agents for any domain, implement A2A communication patterns, and build reliable multi-agent systems that can tackle sophisticated challenges.

The xpander.ai Custom Agent Platform

xpander.ai provides the infrastructure for building custom autonomous agents:

  • Custom Agent Framework - Build agents for any domain or use case
  • Agent-to-Agent Protocol (A2A) - Orchestration and communication between agents
  • Local Functions - Execute custom code and local operations
  • Unified state management - Memory objects to make agents stateful with thread support
  • Tool access - Model Context Protocol (MCP) compatible tool integration
  • Agentic RAG - Optimize API responses for better agent performance
  • Reliable execution - Robust engine for tool use and AI calls
  • Human interfaces - REST API, Web UI, MCP, and webhook support
  • Vault - Secure key management

Workshop Focus: Custom Agents with A2A

This workshop teaches you to build custom agents using coding as a concrete example. The principles apply to any domain:

  1. Custom Agent Design - Design patterns for domain-specific agents
  2. A2A Communication - How agents coordinate and collaborate
  3. Local Functions - Execute custom logic and operations locally
  4. Multi-Agent Orchestration - Manage complex workflows across agents
  5. Practical Implementation - Build coding agents as a real-world example

Workshop Modules

What You’ll Learn

This workshop teaches you:

  • Custom Agent Architecture - Design patterns for building domain-specific agents
  • A2A vs MCP - When to use Agent-to-Agent communication vs. tool access protocols
  • Local Functions - How to execute custom code and operations within agents
  • Multi-Agent Systems - Orchestration, coordination, and communication patterns
  • Practical Implementation - Real-world coding agent examples you can adapt

Key Concepts

Custom Agents

Learn to build agents tailored for specific domains and use cases:

  • Role definition and specialization
  • Custom logic and business rules
  • Domain-specific tools and capabilities
  • Performance optimization for specific tasks

Agent-to-Agent (A2A) Protocol

Understand how agents collaborate:

  • Coordination and task delegation
  • Information sharing between agents
  • Conflict resolution and consensus
  • Workflow orchestration across multiple agents

Local Functions

Execute custom operations within agents:

  • File system operations
  • Database interactions
  • API integrations
  • Custom business logic

Coding Agent Example

While the workshop uses coding agents as examples, the patterns apply to any domain:

  • Repository management agents
  • Code generation specialists
  • Quality assurance agents
  • Project management coordinators

Prerequisites

You need:

  • AWS account with Amazon Bedrock access
  • xpander.ai account
  • GitHub account
  • Basic understanding of AI agents and APIs
  • Programming knowledge (Python/JavaScript helpful)

Beyond Coding Agents

The patterns you’ll learn apply to many domains:

  • Content Creation - Agents for writing, editing, and publishing
  • Data Analysis - Agents for processing, analyzing, and reporting
  • Customer Service - Agents for support, escalation, and resolution
  • Sales & Marketing - Agents for lead generation, nurturing, and conversion
  • Operations - Agents for monitoring, alerting, and automation

Getting Started

Ready to build custom agents that work together? Start with Module 1 to create your first coding agent and learn the foundational patterns for custom agent development.

Go to module 1 to begin building your custom agent system!