Section 1: Foundations, Models & Safe Setup
Before you build any habit, workflow, or automation with Claude, you need to understand what you’re actually working with. Not at a technical level, but at a practical one. What does Claude do well? What does it get wrong? Which model should you choose, and why does that choice matter? How does memory work, and what does Claude actually remember between sessions? And perhaps most importantly: what are your responsibilities as the person reviewing its output?
This section answers all of that. It’s the foundation everything else in the course rests on.
Why this section matters
Most people start using Claude the same way they start using a search engine: they type something in, read the result, and move on. That works fine for casual use. It stops working the moment you start relying on Claude for anything that matters, like drafting communications, summarizing documents, researching decisions, or setting up automated workflows.
The gap between “using Claude” and “using Claude well” comes down to setup and understanding. Learners who skip this section tend to make consistent mistakes: they use the wrong model for the job, they don’t manage memory or context, they share things they shouldn’t, and they trust output they haven’t reviewed. This section closes all of those gaps.
What you’ll learn
This section covers six topics:
What Claude is and how it differs from search. Claude doesn’t retrieve information from the web on demand. It reasons over a large body of trained knowledge to generate responses. That distinction matters for how you ask questions and how much you trust the answers.
Where Claude is strong and where it can fail. Claude is genuinely excellent at drafting, summarizing, explaining, structuring, and reasoning through ambiguous problems. It is less reliable on recent facts, precise numbers, citations, and anything requiring real-time data. Knowing the difference prevents expensive mistakes.
Choosing a model. Claude comes in several variants. Sonnet handles well-scoped tasks efficiently. Opus is built for work that requires interpretation, nuance, and figuring out what you actually want when the brief is vague. Choosing correctly saves time and improves results.
Memory. Claude now has memory on all plan tiers. It auto-summarizes past conversations and carries forward relevant context. You can view, edit, and delete what it remembers in Settings. Understanding this system lets you shape how Claude builds its model of you and your work over time.
Searching and referencing past chats. Your conversation history is searchable. Knowing how to find and reference previous work makes Claude significantly more useful as a long-term thinking partner rather than a one-off tool.
Privacy basics and human review. Claude processes what you send it. There are real implications for sensitive information, confidential business data, and personal details. This topic also introduces the habit of reviewing output before acting on it, which becomes critical later in the course when Claude starts taking actions on your behalf.
The mini project
By the end of this section, you’ll produce a one-page setup guide: your preferred model choices mapped to task types, a note on what you’ve configured in memory and context, a set of personal privacy rules for what you will and won’t share with Claude, and a review checklist you’ll actually use.
This isn’t busywork. The setup guide becomes a reference document you return to, especially when onboarding teammates or revisiting your workflow after a break.
How to approach this section
Read each topic in order. The concepts build on each other. By the end, you should have a clear mental model of what Claude is, a configured workspace, and the beginnings of a review habit.
The goal isn’t to become an expert on how large language models work internally. The goal is to become someone who uses Claude deliberately, catches its mistakes, and doesn’t get caught off guard when it gets something wrong.
That’s a more useful skill than knowing how transformers work.