In this module, you will learn to describe desired workflow behavior in plain language, use AI assistance to propose workflow logic, configure and review that logic in the no-code platform, and refine it without writing conventional code. You will practice turning a natural-language description into working workflow logic, then critically review what gets generated before accepting it. Because generated logic can be wrong or unsafe, you will also practice refining it safely through iterative prompting, catching risky steps before they run. By the end of this module, you will be able to describe desired agent behavior in natural language and responsibly review and refine what an AI tool generates in response.

In this module, you will build two working agents from the ground up. In Lab 1, you will build a calendar/task agent that captures a request, interprets intent, and creates or updates a task or event, asking for approval when needed. You will then extend your skills with a multi-step workflow agent that summarizes information, extracts actions, routes outputs, and sends a follow-up notification. Both builds include a human approval point so the agent doesn't act unsupervised. The module closes with a lesson on testing and troubleshooting, so you can catch problems before they reach a real workflow. By the end of this module, you will have two working, tested agents.

In this module, you will design and build an agent independently, without a guided walkthrough. You will choose a use case, drawing on your own work or personal life, then connect the tools and data sources it requires. You will add guardrails, including permissions, approval steps, and fallback behavior, so the agent handles uncertainty safely rather than acting unsupervised. You will test your agent against common and edge-case scenarios and prepare a short demo of it in action. By the end of this module, you will have an independently built, tested agent ready to finalize as your course project in Module 5.

In this final module, you will finalize your independently built agent into a complete, submission-ready no-code workflow for a real use case, such as a meeting-to-actions agent, a calendar/task organizer, an inbox-to-task workflow, a form-to-summary-and-routing agent, or a recurring status-report assistant. Your submission must include at least one trigger, an AI reasoning step, one or more actions, a human approval or fallback condition, and evidence of testing. The module then closes with a course summary, glossary of key terms, and a final exam covering the course as a whole.