Start from Zero: what Claude is (and what it isn't)

You’ve probably heard that Claude can write your emails, draft your website, and answer almost anything. Some of that is true, some of it is oversold, and almost nobody tells you where the line sits. This guide draws that line — in plain English, before you spend a minute building. Once you know what Claude really is, the rest of this site makes a lot more sense.

What is Claude, in plain terms?

Claude is an AI assistant made by a company called Anthropic. The simplest way to picture it: a sharp, fast, endlessly patient assistant who has read an enormous amount and will help with almost any task you can describe in words — but who you still have to point in the right direction.

You don’t install anything complicated. You open a website — claude.ai — type a message the way you’d text a knowledgeable colleague, and Claude types back. (There are phone and desktop apps too, but the browser is all you need to start.) You can ask it to write something, explain something, look over a document, or build something you can see, like a simple web page.

That’s the whole interface: a conversation. No buttons to master, no manual. If you can describe what you want, you can use Claude.

What is Claude actually good at?

For a small business owner, the genuinely useful strengths are narrower and more practical than the hype suggests:

• Writing and rewriting in plain, human language — an email you’ve been dreading, the text for a web page, a careful reply to an unhappy customer. It’s especially good at making stiff, corporate-sounding text sound like an actual person.
• Explaining things without jargon — paste in a confusing letter, a contract clause, or an error message and ask, “what does this actually mean?”
• Reading long things for you — drop in a multi-page document and ask for the key points, or ask one specific question about it.
• Looking at images — a photo of a handwritten note, a screenshot, a chart. You can show it something, not just describe it.
• Building things you can watch take shape — when you ask for a web page, Claude opens a live preview beside the chat (a feature called Artifacts), so you see the real result, not just code.
• Looking up current information — when you switch on its web search, Claude can check the live web and tell you what it found, with links.

Notice what these have in common: they’re all things you’d hand to a capable assistant. That’s exactly the right mental model.

Here’s a no-stakes way to feel it. Open claude.ai and send this as your very first message:

Read what comes back. That’s the texture of working with Claude.

What can’t Claude do?

This is the part that saves you real frustration later, so here’s the honest version:

• It is not always right. Claude can sound completely confident and still be wrong, or quietly invent a detail that was never true — a fake statistic, a misremembered fact, a quote nobody said. (People call this “hallucinating.”) Treat it as a brilliant first-drafter, never a final authority. Anything that matters — a price, a legal point, a phone number, a claim you’ll publish — you verify yourself.
• It doesn’t automatically know today’s news. Claude’s built-in knowledge stops at a cutoff point (currently around early 2026), so it won’t know something that happened last week unless you turn on web search and let it look. Out of the box, it’s working from memory, not the live internet.
• It can’t see your private world. Claude has no access to your email, your files, your bank, or your accounts. It only knows what you type, paste, or upload into the conversation. That’s a privacy feature, not a flaw — but it means you have to give it the context.
• It’s not a one-click magic button. Claude won’t read your mind. Ask for “a website” and you’ll get something generic; describe your business, your pages, and the feeling you want, and you’ll get something close to right. The quality of what comes back depends almost entirely on how clearly you ask — which is the actual skill this whole site teaches.
• It isn’t a person. There’s no one on the other end to impress or annoy. You can ask the “stupid” question, demand the same thing ten times over, or tell it flatly that it got something wrong. For a beginner, that’s the freeing part: there’s no way to look foolish.

None of these are reasons not to use Claude. They’re the operating instructions. Knowing them is the difference between trusting it blindly and using it well.

Do I have to pay for Claude?

No. You can start completely free at claude.ai with just an email address — no credit card.

And the free plan is not a crippled demo. It’s genuinely capable: it runs on Claude’s fast everyday model and includes the features that matter for getting real work done — the live previews (Artifacts), web search, memory that carries useful context between chats, the ability to upload files, and Projects for keeping a piece of work organized in one place. You can build your entire website on the free plan, following the cornerstone path on this site.

The honest catch is usage limits. Free accounts can do a certain amount of work in a stretch before Claude asks you to wait a few hours and pick back up. On a light day you’ll never notice; on a heavy build day you might bump into it.

The paid plan — Claude Pro, around $20 a month — mainly buys you two things: a lot more usage before you hit a limit, and access to Claude’s most powerful model for harder tasks. It’s worth it once Claude becomes a daily tool. It is not something you need to start.

The sensible move: start free. Only pay when you actually keep hitting the limits — not a day before.

How should I think about working with Claude?

If you take one thing from this guide, take this: treat Claude like a talented assistant you’re directing, not a vending machine you press once.

The whole rhythm is a loop — ask, look, adjust. You ask for something. You look at what comes back. You tell it what to change (“shorter,” “warmer,” “move the contact form to the top”). It revises. You repeat until it’s right. That loop is the entire skill, and it’s identical whether you’re polishing one email or building a five-page website.

That’s what the rest of this site is really teaching — not magic words, just how to direct a very capable assistant toward the thing you actually want.

Ready to point it at your website? The cornerstone path starts here: What you’re building, and why this way.