Your first hour with ChatGPT: a click-by-click walkthrough

Most "ChatGPT tutorials" send you off with a screenshot tour and a wave. Then you sit in front of a blank input box and forget everything. This is a different approach: an hour of structured first use, ordered to get you to a real win as fast as possible.

You will need an email address, a phone for verification, and somewhere quiet to think for sixty minutes.

Minutes 0–5: Sign up and skip the marketing

Go to chat.openai.com. Sign up with a Google account if you have one — it is the fastest path. You will be asked for your name and a phone number. The phone number is for spam prevention, not for billing.

You will then see ChatGPT's home screen. Before you do anything else, do not pay for a subscription. The free tier is fine for an hour. We come back to whether to upgrade in a separate article.

You will see four things on the home screen:

• The big input box at the bottom (where you type).
• A sidebar on the left (your past conversations, hidden on mobile).
• A model selector at the top (which says something like "ChatGPT" or "GPT-4o").
• Your account icon in the top right.

That is the whole product. Everything else is a layer on top of these four.

Minutes 5–10: Send a real first message

Resist the temptation to type "hello." Instead, send something that has a chance of being useful in your actual life. A starter prompt that works for almost anyone:

I want to write a short, professional email declining a meeting invitation politely. The meeting was suggested by a senior colleague at another company. I am not declining because I am busy — I am declining because I do not think the meeting will be useful, but I want to leave the door open for the future. Draft three different versions in three different tones: warm, neutral, and crisp.

Send it. Read what comes back.

Notice three things while reading: how good the draft is for thirty seconds of work, what is slightly off about the tone, and which of the three versions is closest to how you would write.

Minutes 10–20: Have a conversation, not a vending-machine moment

The single biggest mistake first-time users make is sending one prompt, accepting whatever comes back, and either being impressed or disappointed. Instead, follow up.

Reply with something like:

Version 2 is closest. Tighten the opening sentence, drop the second-to-last paragraph entirely, and end with one short sentence instead of two.

Watch what happens. You did not have to re-explain the original email. The model remembers everything you said in this conversation. This is the entire shape of working with AI — it is a back-and-forth, not a search bar.

Try one more turn:

Now rewrite it as if I were declining because of a genuinely full week, but with the same tone.

By the time you have done three or four turns, you will have a usable email and a feel for how the model "thinks" — which is to say, how it responds to corrections.

Minutes 20–30: Try the other useful modes

Look at the icons inside or just below the input box. Their names change occasionally, but as of 2026 the ones you actually want to know about are:

Attach / upload. A paperclip or "+" icon. You can drop in a PDF, image, spreadsheet, or text file. Try uploading a document you have on hand and asking "What are the three decisions someone reading this has to make?" or "Summarize this for a colleague who has five minutes." This is one of the most useful features and most beginners never try it.

Search. A globe icon. ChatGPT does not normally browse the web in real time, but with this enabled it can. Use it for anything where you need current information — news, prices, what was announced this week. Without search on, the model will use its training data, which is months out of date.

Voice. A microphone or headphones icon. Tap, talk, and listen. Great for hands-free use while walking, cooking, or driving. The conversation quality is the same — the only difference is the interface.

Reason / Think. A label like "o1," "GPT-5 Thinking," or "Extended thinking" depending on what model is current. This makes the model slow down and reason more carefully before answering. Use it for tricky logic, math, planning, or analysis. Avoid it for simple chat — it is overkill and slow.

You do not need all four right now. You need to know they exist.

Minutes 30–40: Set up your account properly

Click your account icon in the top right, then Settings. The handful of things worth doing once:

Custom instructions (the "personalization" / "customize ChatGPT" panel — separate from the Memory feature below). Tell ChatGPT who you are and how you want it to respond. Two short paragraphs are plenty. A good template:

About me: I work as a [role] at a [company / industry]. I am based in [country]. I am [interested in / responsible for / learning about] X.

>

How I want you to respond: Be direct, skip the warm preambles, prefer short paragraphs and concrete examples, and call out when you are uncertain.

Save it. From now on, ChatGPT will tailor every reply to that context. You should not have to re-explain who you are every time.

Memory. This is a feature where ChatGPT remembers facts about you across conversations ("I am vegetarian," "my company is in Estonia"). Decide if you want it on. If you have any privacy concerns, leave it off; you can always tell it the relevant facts in a single prompt.

Data controls. This is where you can decide whether your conversations are used to improve the model. Choose what you are comfortable with. If you ever paste in work data, turn this off as a baseline.

Minutes 40–55: Pick a real task and finish it

This is the most important fifteen minutes of the hour. Pick something you already had to do today — a real email, a real summary, a real decision, a real piece of writing — and do it with ChatGPT.

Some choices that almost always work for first-timers:

• Rewrite a draft email so it is half as long but clearer.
• Summarize a long article or document down to "what does this mean for me."
• Plan the next two days of work by pasting your calendar and to-do list and asking the model to help you order them.
• Get unstuck on a piece of writing by describing what you are trying to say in plain English and asking it to draft a paragraph.
• Practice a hard conversation by asking the model to play the other person and push back.

Whichever you pick, finish it. The point is to walk away with one thing that exists because of this hour.

Minutes 55–60: Save what you learned

Scroll back through your hour-long conversation. Notice which prompts worked and which fell flat. The pattern almost always boils down to: more context, more specificity, more constraints.

In the sidebar on the left, click your favorite conversation and rename it ("First hour — email drafts" or similar). You can always come back to it. Open a few of your own threads later in the week to look at how your prompting changes; the improvement curve is steeper than most people expect.

What you now know

You have used the input box, the model selector, file upload, search, voice, custom instructions, and memory. You have had a multi-turn conversation, edited a response, and shipped a real task. Most casual users never get that far.

Tomorrow, pick one task from your normal day and do it through ChatGPT first. The week after that, the habit takes care of itself.