What is Claude? A guide for people who have never used AI

Claude is an AI assistant made by Anthropic. You type a question or a task in plain language, and Claude answers, drafts, summarizes, reads files you attach, or builds documents you can share. You can try it free at claude.ai in a browser, or in the desktop and mobile apps.

I'm Hai Le, a UI/UX designer who codes. I built this site with Next.js and Claude Code, and most of its planning notes were drafted back and forth with Claude. This guide is for people who have never opened an AI tool. Plan names, features and limits come from Anthropic's official pages, checked on 2026-10-04. Prices and limits change often, so reopen the pricing page before you pay for anything.

What is Claude, and who makes it?

Claude is a large language model: software trained on a very large amount of text, which lets it read what you write and respond with something coherent. Anthropic is the company behind it. There are several Claude models, and paid plans let you pick from more of them.

You will run into Claude in a few places:

• claude.ai, the chat window in your browser. Most people start here.
• The Claude apps for desktop and mobile, signed in to the same account.
• Claude Code, which lets Claude work directly with files on your machine. I cover it in installing Claude Code.

This guide is about the first one, Claude as a chat assistant.

How do the free and paid plans differ?

This is a summary of Anthropic's pricing page as of 2026-10-04, in USD.

Plan	Price	What stands out
Free	0	Chat on web, desktop and mobile; web search, file creation and code execution; memory across conversations; app connections; Artifacts
Pro	20/month billed monthly, 17/month billed annually	Everything in Free, plus more usage, Claude Code, Projects, more models and Research
Max	From 100/month	5x or 20x Pro usage, priority at peak times
Team, Enterprise	Per seat	Admin controls, SSO, central billing

Every plan has usage limits. The pricing page describes Free as covering everyday questions, with usage measured in rolling five-hour windows. When you hit the limit, you wait for the next window or upgrade.

My advice: use the free plan for a week or two on real work. Pay for Pro once you keep running out, or once you want Claude Code.

How do I start my first conversation?

1. Go to claude.ai and create an account.
2. Before you paste anything, open Settings, then Privacy, and decide on Help improve our AI models. With it on, Anthropic may train on your chats. With it off, new chats are excluded.
3. Type one small, real task from this week. For example: "I need to move Thursday's design review to Friday. Write a short, polite message to the team."
4. Read the answer, then say what to change: "shorter", "drop the second apology", "a bit warmer". Claude remembers everything earlier in the same conversation, so you don't have to restate the task.
5. Start a new conversation when you switch topics. One topic per conversation keeps answers focused.

Step 4 is where beginners give up. They ask once, the reply is a little off, and they decide AI doesn't work for them. The second or third turn is usually where the result fits.

What can Claude actually do?

Beyond answering questions, five features are worth knowing early.

Files and images. Click the attach button and pick a file. Anthropic's help center lists PDF, DOCX, CSV, TXT, HTML, JSON, XLSX and a few other formats, plus JPEG, PNG, GIF and WebP images.

Chat upload limit	Value (help center, updated July 23, 2026)
Size per file	500 MB
Files per conversation	Up to 20
Image dimensions	Up to 8000 × 8000 px
PDFs	Up to 1000 pages; visual elements are analyzed in the first 100 pages only

For designers, screenshots are the obvious use. Ask "what's wrong with the information hierarchy on this screen?" and you get a list to check against your own judgment.

Artifacts. Anthropic defines an artifact as anything Claude makes that you'd put in front of someone: a document, a deck, a dashboard, a small interactive tool. It opens next to the conversation, you can keep editing it, and you can share it. Artifacts are on every plan, including Free.

Projects. A project is a workspace with its own chat history and its own knowledge base. You upload reference files once and write instructions once, and every chat in that project uses them. Free accounts can create up to five projects. Project files are limited to 30 MB each.

I use projects the way this site's repo uses its AGENTS.md file: anything I would otherwise repeat in every session gets written down in one place. For a design project, that might be the brief, interview notes and a description of the design system.

Research. Claude runs several rounds of web searches, reads what it finds, and returns a cited report within minutes. Research is on paid plans only. Turn it on with the + button at the bottom left of the chat box, then choose Research. Web search has to be enabled.

Memory and connected apps. The pricing page lists memory across conversations and app connections on the free plan. I haven't tested every connector on Free, so I won't list them here.

How do I ask so the answers are useful?

Anthropic's prompting guide has an image that stuck with me: treat Claude like a brilliant new employee who doesn't know your norms yet. It also offers a test. Show your prompt to a colleague with no context and ask whether they could follow it. If they'd be confused, Claude will be too.

A good request usually has four parts:

1. Context: who you are, who the work is for, why it matters.
2. The task: one clear verb, such as summarize, rewrite, list, compare.
3. Material: the text or file Claude needs to read.
4. Output shape: a table, bullets, a length, a tone.

I wrote six copy-paste prompts for everyday design work in how to write prompts for Claude.

One lesson from this site. The mobile menu once opened a language dropdown on top of itself, which was hard to tap. Claude built exactly what I described, and my description was wrong. Since then I write prompts with the care I'd give a design brief for an engineer.

Where does Claude get things wrong?

Claude writes fluently, so wrong answers sound just as confident as right ones. These are the things I always check:

Content	Risk	How to check
Numbers, dates, prices	Outdated or invented	Ask for the source and open it
Software feature names	Products change faster than a model's knowledge	Compare with the official docs
Quotes, book or article titles	May not exist	Search the exact title
Summaries of long documents	Missed or merged points	Ask for verbatim quotes as evidence
Code	Runs but does the wrong thing	Run it, read the diff, ask Claude to explain

The fourth row comes from Anthropic's own guide: for long documents, ask Claude to pull the relevant quotes first and then do the task. You get evidence to check, and the answer stays closer to the source.

Another habit: when Claude reaches a conclusion, ask "what would make this wrong?" The reply often points at the exact thing you should verify yourself.

Should I paste work data into Claude?

That depends on your contract and your company's rules. Before pasting anything, I ask three questions:

• Is this covered by an NDA?
• Does it contain names, emails or phone numbers of real users?
• Does my company have an AI tools policy?

If any answer is yes, I strip identifying details first, for example renaming interviewees to "Participant 1". In my Car From Japan case study, an old screenshot exposed a real email address and phone number, and I flagged it for re-export with sample data before publishing. The same rule applies to anything you hand an AI.

On Anthropic's side, personal accounts have the training toggle described above. Its privacy center also says Incognito chats are not used to improve Claude, even with that toggle on. Team and Enterprise plans are managed by your organization, so ask your IT team.

How is Claude different from Claude Code?

	Claude (claude.ai and apps)	Claude Code
Where you work	A chat window	A project folder, via terminal, desktop app or VS Code
What Claude does	Answers, drafts, builds artifacts	Reads and edits files, runs commands, checks results
Plan	Free plan available	Paid plan required
Good for	Writing, summarizing, ideas, reviewing screenshots	Building pages, editing code, working across a repo

I use both: chat for thinking and writing, Claude Code when a design needs to become a working page. If you're a designer curious about the second, read Claude Code for designers.

What should I read next?

• How to write prompts for Claude: 6 templates for designers, to put Claude to work this week.
• What is Claude Code? Install it and run your first session, when you want Claude working with files.
• What is DESIGN.md, on writing your design system down so AI can read it.
• Learn Figma from zero, if you're new to design.

To see how I work, browse projects like Car From Japan or read about me.

Sources

All checked on 2026-10-04.

• Claude: Plans and pricing: plan names, prices, features per plan, five-hour usage windows.
• Claude Help Center: Upload files to Claude: formats and limits, updated July 23, 2026.
• Claude Help Center: What are artifacts and how do I use them.
• Claude Help Center: What are projects?: five-project limit on Free.
• Claude Help Center: Use Research on Claude: eligible plans and how to turn it on, updated June 2, 2026.
• Anthropic Privacy Center: Is my data used for model training? and How do I change my model improvement privacy settings?.
• Claude Docs: Prompting best practices: the "new employee" framing, the colleague test, quoting before working with long documents.