Differentiate agents from chatbots - You will learn how to tell an AI agent apart from a basic chatbot and choose the right approach for a process, instead of building something too limited or unnecessarily complex.
Design workflows without coding - You will understand the canvas and operating logic of Agent Builder, so you can create task flows, decisions, and data paths on your own without coding or relying on developers.
Break processes into steps - You will learn how to turn a broad business goal into a clear sequence of agent actions, making your solutions more predictable, easier to manage, and simpler to expand over time.
Improve agent prompts - You will practice writing prompts for individual workflow stages, separating system instructions from task context, and fixing common issues that reduce the quality of agent responses.
Build conditional logic - You will master agents that react to different scenarios using if/else logic, exceptions, and alternative paths, so you can support more complex business cases with confidence.
Use documents and data - You will learn how to feed an agent with user inputs and content from PDF, DOCX, and text files, so it can analyze information, make decisions, and act in real context.
Test and optimize your agent - You will learn scenario testing, debugging, and versioning methods that help you catch errors faster, reduce response time, and control the operating cost of your solution.
Prepare for company rollout - You will understand how to deploy agents for employees and customers, assign responsibilities in the team, and maintain quality, security, and governance across the organization.
AI agents are positioned here against chatbots to clarify what makes them distinct in practice. The section follows the progression from prompt to workflow to agent, reviews company use cases, and indicates situations where adopting agents is not the right choice.
ChatGPT AI Agent Builder is presented through its operating model, showing how canvas and workflow structures define an agent. The section explains nodes for input and output, LLM tasks, decisions, conditions, tools, integrations, and the way data moves through the flow.
Designing agent behavior starts with process thinking rather than isolated prompts. Participants break work into steps, then map sequential and decision-making logic for cases such as handling customer inquiries or analyzing a document before taking action.
Prompt design in agents requires a different approach than writing a single standalone instruction. The section distinguishes system instructions from task context, explains context management, and analyzes common prompt mistakes through response-improvement practice.
When one agent must react differently across situations, logical branching becomes essential. This section uses if / else paths, exception handling, and error management to build multi-step behavior, illustrated through practical HR and offer-agent exercises.
Working with PDFs, DOCX files, and plain text is covered together with the use of user-provided input inside agent flows. Attention is also given to tool integrations in the MCP model, as well as clear boundaries for data access and security.
Reliable operation depends on structured scenario testing and careful workflow debugging. The section also addresses how to reduce response time and operating cost without weakening results, and how versioning supports controlled changes to an agent over time.
From an organizational perspective, the section focuses on where agents can be launched and how deployment differs for employee-facing and client-facing use cases. It also clarifies ownership, operating responsibilities, governance rules, and the controls needed to maintain output quality.
Basic AI tool usage - You should be comfortable using ChatGPT or similar tools and know how to phrase prompts, because the training builds on AI usage at the level of processes and agents.
Process thinking - You should be able to describe a simple process step by step, identify decisions and exceptions, and understand links between stages, because agent design depends on it.
Working with documents and data - You should be comfortable handling documents and input data, understand their role in a task, and judge what information is needed before a decision can be made.
Business environment awareness - You should understand how processes work in your company or team, so you can design realistic rollout scenarios, testing approaches, and clear roles and responsibilities.
