Hands-on Activities

Everything you need for each week’s in-class hands-on and exploration work lives here: the flow, the data files, and the exact prompts to copy into your agent. One page per week.

You will still do some of the work on an external platform — prompting ChatGPT or Claude, tidying in Google Sheets, building dashboards in Looker Studio. Those steps are marked clearly on each page. This site is your single starting point; it does not run anything itself.

How each activity page is laid out

1. At a glance — timing, who you work with, what you hand in.
2. The scenario — the business situation behind the data.
3. The steps — what to do, in order, with copy-ready prompt blocks.
4. Your data files — download links for this week’s datasets.
5. Verify & submit — the checks to run and what to put in your portfolio.

This week

Week	Session
2	Directing and Supervising AI Agents	Open now →

Each week’s activity opens here when we reach it in class — check back at the start of the session.

Working with an agent — the habit this module builds

Across every activity the pattern is the same, and it is the thing you are actually being assessed on:

1. You look first. Open the data yourself before any agent sees it. Write down what you notice.
2. You brief the agent. Give it your plan, not a blank instruction — and tell it to report before and after.
3. You verify by hand. Trace at least one number back to the source. A result you did not check is not evidence.
4. You decide. Accept, change, or reject each thing the agent did, and say why.

Speed is not graded. Traceability is.