Software development
Artificial Intelligence
Cloud
View all courses
Learn
Connect
Blog
Resource hub
Featured resource
2025 Tech Upskilling Playbook
Tech Upskilling Playbook

Build future-ready tech teams and hit key business milestones with seven proven plays from industry leaders.

Check it out

Build Multi-Agent Systems with OpenAI Agents SDK

Course Summary

This course transitions developers from building linear, single-prompt LLM applications to engineering dynamic, multi-agent swarms. Utilizing the OpenAI Agents SDK, participants will master the Handoff architecture, where specialized agents autonomously transfer control to one another based on task requirements. The course emphasizes Durable Execution (via Temporal) and Persistent Sessions, ensuring agents can handle long-running, complex workflows that survive system restarts. From building "Triage Agents" that route user intent to deploying Model Context Protocol (MCP) servers for universal tooling, this course provides a production-ready blueprint for high-reliability AI orchestration.

Prerequisites:

In order to succeed in this course, you will need:

  • Intermediate to advanced Python
  • API & Web Fundamentals including OpenAI API and REST API experience
  • Basic understanding of agentic workflows
  • An understanding of CI/CD pipelines and containerization
Purpose
Transition from building linear, single-prompt LLM applications to multi-agent swarms using OpenAI Agents SDK
Audience
Senior level programmers and engineers looking to move beyond single-prompt LLM applications
Role
AI Engineers & LLM Developers | DevOps & Platform Engineers | Software Architects | Backend Developers
Skill level
Intermediate
Style
Lecture | Hands-on Activities | Labs
Duration
3 days
Related technologies
OpenAI | Gen AI | Temporal 

 

Learning objectives
  • Implement Agent Handoffs
  • Manage Stateful Sessions
  • Engineer Type-Safe Tools
  • Architect Durable Workflows
  • Standardize External Access using the Model Context Protocol (MCP)

What you'll learn:

In this Build Multi-Agent Systems with OpenAI Agents SDK course, you'll learn:

The Agentic Loop & Handoff Primitives

  • Transitioning from Swarm and Assistants API to openai-agents SDK
  • Core Concepts
    • Agent
    • Runner
    • Think-Act-Observe Loop
  • The Handoff Pattern - transferring control from one agent to another specialized agent
    • Passing state and user data during handoff

Advanced Tooling & Persistent Sessions

  • Configuring Session objects to move past stateless API calls
  • Stateful Runners - Function Tooling for defining schemas
    • Using Pydantic
    • The @function_tool decorator
  • Session Management - maintaining conversation history automatically
    • SQLiteSession
    • PostgresSession
  • Model Context Protocol (MCP) - connecting OpenAI agents to external tool servers
  • Agent-as-Tool - nesting an agent inside another agent’s toolset using .agent_as_tool()

Production, Safety, and Durable Execution

  • Guardrails within the agent workflow
    • Input/output validation
    • Safety “graders”
  • Durable Agents - integrating with Temporal for long-running agents
  • Human-in-the-Loop (HITL) - adding manual approval steps for high-stakes tool calls
  • Observability & Tracing
    • Using the OpenAI Traces dashboard
    • Debugging complex agent-to-agent reasoning chains

Build Multi-Agent Systems with OpenAI Agents SDK

real world content

Real-World Content

Project-focused demos and labs using your tool stack and environment, not some canned "training room" lab.
expert practitioners

Expert Practitioners

Industry experts that bring their battle scars into the classroom.
experiential learning

Experiential Learning

More coding than lecture, coupled with architectural and design discussions.
fully customized

Tailored Outlines

Once-size-fits-all doesn't apply to training teams. That's where we come in!

Dive in and learn more

When transforming your workforce, it’s important to have expert advice and tailored solutions. We can help. Tell us your unique needs and we'll explore ways to address them.

Let's chat

By clicking submit, you agree to our Privacy Policy and Terms of Use, and consent to receive marketing emails from Pluralsight.
Related Pages