Getting Started With ChatGPT, Claude and Gemini

If you have never used an AI assistant, start here. In ten minutes you will have an account, know what you can do for free and have had your first useful conversation. Everything here applies to the three most used assistants: ChatGPT (by OpenAI), Claude (by Anthropic) and Gemini (by Google).

Pick an assistant (and don’t overthink it)

All three do the same essentials: chat, write, summarise, explain, translate, analyse files and search the web. They differ in style, some features and how they integrate with other tools:

Assistant	Company	Free	Entry paid plan (US)
ChatGPT	OpenAI	Yes	Go $8 · Plus $20 a month
Claude	Anthropic	Yes	Pro $20 a month
Gemini	Google	Yes	Varies by country

If you live in Gmail and Google Docs, Gemini fits naturally; if you work with long texts or code, people tend to like Claude; ChatGPT is the most widespread and has the broadest set of features. But to learn, pick whichever you like: what you learn works in all three. Current prices are in our model comparison.

Create your account

1. Go to the official website (chatgpt.com, claude.ai or gemini.google.com) or download the official app from your phone’s store. Beware of look-alike apps that charge for what is free.
2. Sign up with your email or your Google or Apple account.
3. Start on the free plan. You can always upgrade later.

Before you type: check privacy

Almost nobody does this, and it is worth doing on day one. All three companies may use your conversations to improve their models if you allow it:

• ChatGPT: under Settings → Data controls, “Improve the model for everyone” decides whether your new chats are used for training. If you turn it off, they still appear in your history but are not used for training.
• Claude: under Settings → Privacy, the “Help improve Claude” setting. If you allow it, Anthropic may use your chats for training and keep them for up to five years; if not, standard retention is 30 days.
• Gemini: “Gemini Apps Activity” is on by default for users aged 18 or over. You can turn it off or delete activity under Settings & help → Activity.

There is no “right” option: decide knowingly. And whatever you choose, follow the golden rule from our hallucinations and privacy guide: don’t share anything you wouldn’t publish.

Your first conversation

Forget magic words. AI works best when you talk to it like a very capable person who has just arrived and knows nothing about you. Compare:

❌ “Write an email to my boss.”

✅ “Write a short, friendly email to my manager asking for Friday off for a medical appointment. I’ve been at the company three years and have never asked for single days off. Offer to finish Thursday’s report beforehand.”

The second request has a task, context and what you expect. That changes the answer completely. Our prompting techniques guide has seven more.

Keep talking: the second answer is the good one

An AI assistant is not a search engine: you can keep talking. If the answer isn’t right, don’t start over; correct it:

• “Shorter, in three sentences.”
• “More formal, it’s for a client.”
• “I don’t get step two; explain it with an example.”
• “Where does that figure come from? Give me the source.”

That back-and-forth is the most useful skill you can build.

What to know from day one

• It can be confidently wrong. It sometimes invents facts, quotes or sources. Check what matters, especially figures, dates, health, money and legal topics. See why AI makes things up.
• It doesn’t know what happened yesterday unless it searches the web. Its knowledge has a cutoff date.
• Each conversation starts from zero unless you turn on memory or use projects. See projects, instructions and memory.
• To understand how it works inside, continue with how language models work.