Modern ChatGPT Workspaces and Multimodal Tools By the end of this lesson, students should be able to: Automation changes the risk profile because an error can repeat. A safe automation specifies: what to do, when to do it, what sources to use, when to notify, what not to do, and when human approval is required. A reminder simply returns an instruction later. A recurring task produces scheduled output. A condition watch checks whether something has changed. Agentic work may research, create artifacts, and interact with connected tools. Approval gates are essential before sending messages, spending money, changing production systems, publishing public content, deleting data, or making commitments. Use read-only access where possible. Start with one manual run, then a limited pilot, then recurring execution. Review logs and outputs before increasing scope. No-news behavior: condition watches should say not to notify when nothing meaningful changed. Otherwise users receive noise and stop paying attention. Failure behavior: specify what the system should do when a source is unavailable, credentials expire, or evidence conflicts. Automation should reduce repetitive effort, not hide responsibility. The accountable human still owns the goal, permissions, review, and consequences. Enroll to continue this lesson. The preview above shows the lesson objectives and opening lesson content. Enroll to view the full lesson, complete the practice work, and take the lesson quiz.
Lesson 4: Tasks, Work, and Supervised Automation
Lesson Objectives
Lesson Content
