OpenAI Agent Builder Workflows

What you'll learn

• Explain where a visual agent workflow builder fits next to the Agents SDK, chat UI embedding and evaluation tools.
• Design a workflow from start, agent, tool, logic and end nodes with clear state and structured outputs.
• Route requests with if/else branches, bounded loops, handoffs and human approval steps.
• Add guardrails, least-privilege tools and defences against prompt injection.
• Test workflows with previews, traces and evaluation datasets before publishing a version.
• Estimate cost per run and operate published workflows with versioning and monitoring.

Syllabus

What Agent Builder Is and When to Use It

1. Agent Builder in the OpenAI Platform
2. Visual Builder or Code?
3. The Workflow Mental Model: Nodes, Edges and State

Core Building Blocks

1. The Start Node: Inputs and State Variables
2. Agent Nodes: Instructions, Model and Output Format
3. Tools: File Search, MCP and Functions
4. End Nodes and Workflow Outputs

Logic: Branches, Loops and Approvals

1. If/Else Routing on Structured Fields
2. Loops With Exit Conditions and Caps
3. Human Approval Steps

Data, State and Templating

1. Set State and Transform Nodes
2. Structured Outputs Between Nodes
3. Templating Instructions With Variables

Multi-Agent Patterns

1. Triage and Specialist Agents
2. Handoffs, Agents as Tools and Fixed Sequences
3. Keeping Agents Narrow

Guardrails and Safety

1. Guardrail Nodes
2. Least Privilege for Tools and MCP Servers
3. Prompt Injection Through Files, Web and Tools

Testing, Evals, Versions and Deployment

1. Preview Runs and Traces
2. Evaluating Workflows With Datasets
3. Publishing and Versioning Workflows
4. Deploying: Embedded Chat or Exported Code

Cost, a Case Study and a Checklist

1. Estimating Cost and Latency per Run
2. Case Study: A Refund Assistant
3. An Agent Workflow Checklist