What Claude is and where you can use it

Claude is an AI assistant made by Anthropic. You write to it in plain language and it writes back: answers, drafts, summaries, code, explanations, outlines. It is a language model, which means it predicts useful text based on patterns it learned in training and on whatever you put in front of it during the conversation.

This club is an independent community. We are not affiliated with Anthropic, and nothing here is official guidance from them. When we say something about product details, we point you to their documentation so you can check it yourself.

What it is good at

Claude is strong at work that involves language and structure. Reading a long document and pulling out the key points. Turning rough notes into a clean email. Explaining a concept at the level you ask for. Writing and reviewing code. Brainstorming options and then helping you compare them.

What it is not

It is not a search engine and it is not a person. It can be confidently wrong. It does not remember you between separate conversations unless a feature you turned on says it does. It only knows what it learned in training plus what you give it in the current conversation. Treat its output as a strong first draft that you check, especially for facts, numbers, names, and anything with legal, medical, or financial weight.

The main ways to use it

There are three families of surfaces. Names and plans change, so check Anthropic's site for the current list.

• The Claude apps. Chat in a browser, on desktop, or on your phone. This is where almost everyone starts, and where Track A lives.
• Claude Code. A tool that works inside your terminal and your project folder, reading and editing files with your permission. Track B covers it.
• The API. For developers who want to build Claude into their own software. Track C covers the first steps.

All three talk to the same family of models. The difference is how you reach them and what they can touch.

A few words you will see

• Prompt: what you send to the model.
• Context: everything the model can see right now, including your messages, its replies, and any files you attached.
• Token: the small chunks of text a model reads and writes. Limits and pricing are usually measured in tokens.
• Hallucination: a fluent answer that is simply not true.

How to think about it

A useful mental model: Claude is a very fast, well-read colleague who has not met you, does not know your situation, and will happily guess when you leave gaps. The more clearly you describe the situation, the better the result. That is the whole subject of the next lesson.

Try it

Open Claude in whichever app you prefer and run this three-step experiment.

1. Ask: "Explain what a mortgage escrow account is in three sentences."
2. Ask again: "Now explain it to a ten-year-old, in two sentences."
3. Ask: "What would you want to double-check in your own answer?"

Notice how the same question changes shape with the audience, and how the model can point at its own weak spots when you ask. Save the conversation. You will reuse it in the capstone.

Further reading: the official documentation at docs.claude.com and the product pages at anthropic.com.