Week 1 — Agentic Cowork Tools

Welcome to Week 1! Week 1 is open to everyone. The goal: by the end of the week you can research, report, and build scientific software noticeably faster with Claude, Claude Cowork, and Claude Code — anchored in a working understanding of how LLMs and agents actually behave, so you can validate what they produce.

Driver's Ed, then build the vehicle

The mental model for these two summer weeks is learning to drive. Agentic tools can get you from Point A to Point B very quickly, so Week 1 is Driver's Ed: learn to steer them safely and get where you're going. You'll cover how the tools work (the basic mechanics and rules of the road), how to use them (where the accelerator and the brakes are), when to slow down and use caution, and how to navigate and share the routes you find.

Week 2 is where you build the vehicle — assembling the components into your own agentic workflow: a robust utility vehicle (think a Hilux or Tacoma, not a Ferrari) that keeps going whatever you throw at it. The best way through both is to learn by doing. To drive a car, you have to drive a car, and these tools have made learning-by-doing easier than it has ever been — so experiment.

Objectives

By the end of the week you should be able to:

1. Explain how LLMs work — not as black boxes, but understanding why they succeed and fail.
2. Understand the agent paradigm — models that use tools, operate in loops, and interact with your environment.
3. Write effective context — give agents the domain knowledge they need for scientific work.
4. Use Cowork and Claude Code — practical tools for data analysis, code development, and documents.
5. Complete a hands-on project — apply these tools to a real task in your own work area.

The goal is not to make you an AI expert. The goal is accurate mental models and enough practice to decide where these tools are useful — and where they're not.

The week at a glance

• Day 1 — How LLMs & agents work. The mechanics you need to drive well.
• Day 2 — Meeting our vehicles. Getting hands-on with the tools we'll use.
• Day 3 — Skills & tools. Advanced navigation.
• Day 4 — Claude Code & ethics. Driving techniques and what to watch for.
• Day 5 — Practice. Time behind the wheel.

Project time

Every day includes dedicated project and office-hours time from 1–2 PM so you can apply these tools to something you actually care about:

• Bring your own problem — a dataset that needs analysis, a script that needs debugging, a report that needs writing, or a repetitive task you'd love to automate from your real work.
• Work in teams — this is best done together, so find a partner or two.
• Not graded — it's a structured way to leave with practical experience, not just theory.

The week's core

Skills give agents superpowers, and they are the most effective means of knowledge transfer: they let organizations scale by packaging and distributing capabilities. They can also be embedded across every agentic system — including the Cowork tools this week and the agentic systems we build the following week.