How to Use Claude: A Practical Guide for Beginners

Learn how to use Claude from scratch: sign up, choose a plan, write good requests (prompts), work with files and Projects and protect your data.

Claude is an AI assistant developed by Anthropic. You write to it in plain language, like a colleague, and it replies: it drafts texts, summarises documents, explains concepts, analyses data and helps structure ideas. This guide is independent: TecLab is not affiliated with Anthropic, and Claude is a trademark of Anthropic.

You do not need to code. In this guide we explain how to use Claude from scratch: signing up, choosing a plan, writing good requests (known as prompts), working with files and protecting your data.

What Claude does well (and where it fails)

• Drafting and rewriting: emails, proposals, reports.
• Summarising and comparing: long documents, minutes, contracts (preliminary analysis).
• Explaining and studying: new concepts, step by step.
• Analysing data: tables and numbers, with simple conclusions.

Claude can also be wrong. It may present incorrect or invented information as fact, and it does not know your company's context unless you provide it. Treat answers as a first draft and verify facts, figures and sources before using them.

How to use Claude: first steps and choosing a plan

1. Sign up. Use claude.ai or install the app for computer or phone.
2. Start with the free plan. It is enough to try it and see whether it suits you.
3. Write your first request. For example: "Explain in three sentences what an annual budget is."
4. Refine. Reply as in a conversation: "shorter", "more formal", "give me an example".
5. Try a real request from your day, without sensitive data.

Which plan to choose. There are five plans. Free is for trying it out. Pro is for frequent use. Max is for heavy use, with more capacity. Team is for teams, with central administration. Enterprise is for organisations with security and compliance requirements.

All have usage limits, which reset in five-hour windows, and paid plans also have weekly limits. Prices and features change, so always check the official plans page.

How to write good requests (prompts) for Claude

Anthropic advises treating Claude like a capable new colleague with no context: the clearer the request, the better the answer, according to the official guide. Five elements always help:

• Context: who you are, what the text is for, who will read it.
• Goal: what exactly you want to get.
• Format: length, structure, list or running text.
• Tone: formal, friendly, direct.
• Examples: a text of yours that shows the style you want.

Weak request: "Write an email to a customer."

Better request: "I run an accounting SME. Write a 100-word email, in British English, to a client who is late sending documents. Friendly but firm tone, with a clear deadline at the end."

If information is missing, ask Claude to ask you questions before it starts, and do not accept the first answer: iterate.

Files, documents and Projects

Files. You can attach PDFs, Word documents, spreadsheets and images and ask for summaries or comparisons. Supported types and limits are in the official help. Before attaching, check whether the document has personal or confidential data.

Projects. A Project is a workspace with its own instructions and knowledge files, where all conversations share the same context. According to the official help, it is available to all users, and the free plan allows up to five. A good example is a "Sales proposals" project with your style guide and previous proposals.

Going further

• Web search: answers with recent information, when enabled.
• Memory: saves topics across conversations; you can view, edit, delete or turn it off.
• Connectors: link Claude to apps you already use.
• Artifacts: documents, dashboards and interactive pages that appear beside the conversation and that you can keep editing, according to Anthropic's tutorial.

Availability of each feature varies by plan.

Claude and privacy: how to use it safely

On the Free, Pro and Max plans, you can choose whether your new conversations may be used to improve the models. According to Anthropic, if you decline, 30-day retention continues, and if you accept, data may be kept for up to five years. You can change the choice in privacy settings. The Team and Enterprise plans are governed by commercial terms, outside this change.

There is also incognito mode, where the conversation is not saved to history or memory. But the golden rule is simple: do not paste what you would not put in an external email. Customers' personal data, credentials and trade secrets stay out. For work, use a business account and read GDPR and Artificial Intelligence and Shadow AI.

Good practice and questions to ask

Good practice comes down to four things: give context, split big tasks into steps, always verify figures, dates and quotes, and keep your best requests as templates. The most common mistake is treating it as an infallible search engine.

Before using it at work, answer these questions:

• Does my company allow the use of AI, and with which account?
• What kind of data will I paste? Is it personal, customer or confidential?
• Does my account use conversations for training? Can I change that?
• Who reviews the output before it goes to a customer?
• Is there an AI usage policy in the company?
• Do I know how to save and reuse the requests that work?

Frequently asked questions

Is Claude free?

There is a free plan, with usage limits. Paid plans give more capacity and features; see the official page.

Does Claude work in Portuguese?

Yes. Anthropic says Claude works in many languages, including Portuguese. For European Portuguese, ask for it explicitly ("português de Portugal") and review the text.

Does Claude keep my conversations?

It keeps your history and, if memory is on, topics for future context. You can view, edit, delete or turn off memory and use incognito mode. Use for training depends on your plan and your choice.

Can I use it at work with customer data?

Only with company authorisation, on a suitable account and respecting the GDPR. If in doubt, do not paste personal data.

Can I trust the answers?

As a starting point, yes. As final truth, no: always verify what matters.

Conclusion: start small, with clear rules

Claude pays off most with clear requests, good context and human review. Start with a simple task, without sensitive data, and build confidence gradually.

At TecLab, we support Portuguese SMEs in adopting AI, from first use to integration into processes. To put AI to work in your team with clear rules, talk to us.

Read also: Shadow AI: the Invisible Risk of the AI Tools Your Team Already Uses, How to Implement AI in Your SME and GDPR and Artificial Intelligence.