The seven things you can ask ChatGPT on day one

The hardest part of starting with ChatGPT is the blank input box. You sit there and feel slightly silly. You type "hello," it says "hello," and you wonder what the fuss is about.

The fix is to skip the small talk and ask for something specific. The seven prompts below cover the use cases first-time users almost always under-explore. Each one is copyable. Edit the bracketed bits to fit your life and send.

Do all seven in one afternoon and you will have a surprisingly deep feel for the tool — better than many people who have used it casually for a year.

1. Summarize something long into something useful

Find any document longer than a page — a contract, a long email thread, an article you saved, a PDF report, a meeting transcript. Paste it in or upload it, then send:

Read this carefully. Give me back:

>

1. Three sentences someone in a hurry needs to know. 2. The three decisions a reader has to make. 3. Anything that contradicts itself or sounds suspicious.

>

Use plain language. Mark anything you are unsure about with [unclear].

This single prompt is more useful than a hundred "summarize this" requests. You are telling the model what kind of summary you want. The [unclear] tag at the end is the magic ingredient — it gives the model permission to flag uncertainty instead of making something up.

2. Rewrite a piece of your own writing

Take any email, message, or paragraph you wrote in the last week and paste it in with:

Here is something I wrote: [paste].

>

Rewrite it three times:

>

- Half as long, same meaning. - In a warmer tone, no longer than the original. - In a sharper, more direct tone.

>

Then tell me which version you would send and why.

You will learn two things instantly: how to ask for variants instead of a single answer, and how compressing without losing meaning is something the model does well.

The "tell me why" part is worth keeping. It forces the model to compare its own outputs, which is a sneaky way to get a better recommendation than asking it to pick from the start.

3. Plan something you have been putting off

Anything from a trip to a kitchen renovation to a difficult conversation. Try:

I want to [plan a four-day trip to Lisbon in early September / move into a new flat next month / have a hard conversation with my manager about workload]. I have not started.

>

Ask me five questions you would need answered to help me plan this well. Wait for my answers before you continue.

Notice the last line. Most people send an open-ended planning prompt and get a generic answer. By making the model interview you first, you force it to use your actual situation — and you end up with a plan that fits.

4. Explain something three different ways

Pick something you sort-of-but-not-really understand. A line from a contract. A medical term from your last appointment. A piece of jargon from your industry. Send:

Explain [this term / this paragraph] three different ways:

>

- To someone with no background at all. - To a working professional in a different field. - To someone who has the basics but is rusty.

>

Use a short concrete example in each.

By the third version, the concept usually clicks. This works for chemistry, code, contract law, mortgage terms, philosophy, sports rules — anything dense.

5. Quiz yourself on something you are learning

This is the prompt most beginners underrate the longest. If you are trying to learn a new topic — a new role, a certification, a language, a software tool — try:

I am learning [topic]. Quiz me as if I am a smart beginner. Start with easy questions and get harder. Give me one question at a time. Wait for my answer before grading or moving on. After ten questions, tell me which areas I should focus on.

Active recall is one of the best-studied ways to learn anything, and it is awkward to do alone. The model is happy to play the teacher's role indefinitely.

A small refinement: ask the model to not give you the answer if you write "skip." That lets you blow past easy questions without breaking the flow.

6. Draft a hard message you have been avoiding

A complaint to a service provider. A polite chase on an overdue invoice. A note declining an invitation you do not want to accept. A reply to a passive-aggressive email. Try:

I need to write [a polite but firm complaint to my landlord about the heating, which has been broken for two weeks despite three messages].

>

Draft three versions:

>

- Calm and matter-of-fact. - Slightly firmer, with a deadline. - Final-warning tone, but still professional.

>

Avoid being sarcastic or aggressive. End each one with a clear next step.

The reason this works is that the hard part of these messages is not the words — it is overcoming the inertia of writing them at all. Once you have three drafts, picking and editing one takes a minute.

7. Be a thinking partner, not an answer machine

This is the most general-purpose prompt of the seven, and the one that turns ChatGPT from a tool into a habit. Pick any choice you are wrestling with. Big ones, small ones. Try:

I am deciding whether to [take the new job offer / replace my old laptop now or wait six months / start running again three times a week].

>

Do not give me your opinion yet. Instead:

>

1. Ask me four questions to understand my situation better. 2. Once I answer, list the three strongest arguments on each side. 3. Tell me what evidence would change the answer. 4. Then, only after all of that, give me your honest recommendation.

Notice the structure. You are forcing the model to slow down. Most people pose a decision and the model picks one within fifteen words. With this scaffolding, you get something that actually helps you think.

This pattern works for career choices, business decisions, relationship questions, financial trade-offs — anywhere being smart about the question matters more than getting an answer fast.

What ties these seven together

Look at the prompts side by side. They share four habits:

1. They give the model a job, not a topic. "Plan a trip" is a topic. "Ask me five questions you would need answered before planning my trip" is a job.
2. They ask for variants. Three versions, three tones, three explanations. Comparing options is more useful than picking the first one.
3. They include a constraint. "Three sentences." "No sarcasm." "Wait for my answer." Constraints lift quality more than length does.
4. They allow uncertainty. [unclear] tags, "skip" commands, "tell me what would change your answer" — none of which sound like normal prompts, all of which produce better output.

If you do all seven this week, you will not need anyone's "100 ChatGPT prompts" list. The patterns you take from these will generalize to anything else you want to ask.