10 Prompts to Try Right Now

By the end of this lesson you will have run ten real prompts through an AI chatbot and seen what each one gives you back. You will also know how to tweak any of them so the answer fits your life instead of some imaginary average person. Nothing here takes longer than two minutes to try.

The idea in plain English

A prompt is just what you type into the box. That’s the whole word. When people say “prompt engineering,” they mean getting better at typing that request, and it is much less technical than it sounds.

Here is the thing most beginners get wrong: they treat the AI like a search engine. They type three words, get something bland, and decide AI is overrated. But a chatbot like ChatGPT, Claude, or Gemini is not looking things up in a list. It is writing you a response, one piece at a time, based on what you gave it. So what you give it matters enormously. Three words gets you a three-word-quality answer.

The good news is that you don’t need a secret formula. You need to be specific in the same way you would be specific with a competent new assistant on their first day. That assistant doesn’t know your job title, who your customers are, or how formal your emails usually sound. Tell them, and the work comes back usable.

The ten prompts below are working examples of that. Some are useful on their own. All of them are patterns you can reuse with your own details swapped in.

How it actually works

Almost every prompt that works well contains some mix of four things. You do not need all four every time, but noticing them makes it much easier to fix a bad answer.

1. The task. The actual verb. Summarize, rewrite, explain, compare, brainstorm, draft, critique. Vague verbs like “help me with” produce vague results. “Rewrite this to be two sentences shorter and less stiff” produces something you can use.

2. The context. Who you are, who this is for, what the situation is. “I’m a bookkeeper writing to a client who is 40 days late paying” changes the output far more than any clever phrasing.

3. The format. How you want it delivered. A bulleted list, a table, three options, under 100 words, an email with a subject line. If you don’t say, the AI guesses, and it usually guesses “five paragraphs of prose.”

4. The example or constraint. Anything that narrows the target. “Sound like this: [paste something you wrote].” Or “no exclamation points.” Or “assume the reader has never used a spreadsheet.”

One more mechanic worth knowing before you start: the conversation has a memory within itself. Everything you and the AI have said in that chat window stays in view, so you can just say “shorter” or “less formal” or “now do the same thing for my other product” and it knows what you mean. Beginners often start a brand new chat for every attempt and lose all that setup. Stay in the thread and refine.

Try this now

Open ChatGPT, Claude, or Gemini — any of the free versions is fine. Work through these in order. Each one has a bracketed part for you to replace with your own details; delete the brackets when you do.

1. Explain something you’ve been quietly confused about. Copy this in and fill the blank:

   Explain [compound interest] to me like I'm a smart adult who just never learned this. Use one everyday analogy, keep it under 150 words, and skip the jargon. Then give me one number example with real figures.

2. Now push back on it. In the same chat, type: “That analogy didn’t land for me. Try a different one.” Notice that you didn’t have to re-explain anything. That’s the conversation memory doing its job.
3. Rewrite something awkward. Find a real email, text, or paragraph you wrote and weren’t happy with. Paste it after this prompt:

   Rewrite the message below so it's warmer and about 30% shorter. Keep my meaning exactly the same. Do not add any new promises or details I didn't write. Give me two versions: one slightly formal, one casual.
   
   Here's my message:
   [paste your message here]

4. Summarize something long. Paste in an article, a long email thread, or meeting notes and ask: “Summarize this in five bullets, then tell me the one thing I need to decide.” That second half is the part people forget, and it’s usually the useful part.
5. Make a decision easier. Try: “I’m choosing between [option A] and [option B] for [situation]. Give me a table comparing them on cost, time, and risk. Then tell me which you’d pick and why.” The table is what makes this readable instead of a wall of hedging.
6. Draft the thing you’ve been avoiding. The blank page is the hardest part, and AI is genuinely good at getting rid of it. “Write a first draft of [a thank-you note to a client who referred someone / a Facebook post announcing my new hours / a polite decline to a wedding invitation]. Keep it to 80 words, plain language, no exclamation points.” You will rewrite parts of it. That’s fine — you’re editing instead of staring.
7. Ask it to interview you. This one surprises people:

   I want to [write a short bio for my website]. Instead of writing it yet, ask me five questions one at a time that will help you write it well. Wait for my answer before asking the next one.

   The “one at a time” and “wait for my answer” bits matter — without them you get all five questions dumped at once, which is much harder to answer.

8. Brainstorm wide, then narrow. “Give me 15 ideas for [low-cost ways to thank my regular customers]. Make them varied — don’t give me 15 versions of the same idea.” Then pick three and say: “Expand these three into something I could actually do this week.” Quantity first, quality second. Asking for one perfect idea gets you a mediocre one.
9. Have it check your work. Paste in anything you’re about to send and ask: “Read this as if you were [my skeptical boss / a customer who’s already annoyed]. What would they misread, object to, or find unclear?” This catches things spellcheck never will.
10. Turn a mess into a plan. Dump your actual to-do list in, unsorted and messy, and ask: “Here’s everything on my plate. Group these into themes, flag anything that looks like it depends on something else finishing first, and tell me what to do tomorrow morning.” The AI does not need your list to be tidy. That’s the point.

If you only do three of these, do numbers 1, 3, and 7. Number 7 in particular tends to be the one that changes how people use these tools.

Common mistakes

• Typing too little and blaming the tool. “Write me a marketing email” will give you generic filler, because you gave it nothing to work with. Add who it’s for, what you sell, and how long it should be. Twenty extra words of context typically doubles the quality of the answer.
• Accepting the first draft. The first answer is a starting point, not a verdict. Say “shorter,” “less corporate,” “you missed the main risk,” “try again but assume the reader is skeptical.” Two or three rounds of that gets you somewhere good. People who find AI useful are almost always people who reply to it.
• Starting a new chat for every attempt. If you open a fresh window each time, you throw away all the context you built up and start from zero. Refine inside the same conversation, and only start a new chat when you genuinely switch topics.
• Trusting facts, numbers, and quotes without checking. These tools can state wrong things in a completely confident tone — this is often called hallucination. It is fine for drafting, rewriting, brainstorming, and explaining. For a statistic, a price, a legal detail, a person’s title, or a citation, verify it somewhere else. Perplexity is useful here because it shows the sources it used.
• Pasting in things you shouldn’t. Don’t paste client bank details, passwords, medical records, or anything covered by a confidentiality agreement into a free consumer chatbot. Change the names, remove the numbers, or describe the situation generically. Your prompt does not need real identifying details to be useful.

Key takeaways

• A prompt is just what you type. Getting better at it is mostly about being specific, not learning tricks.
• Most strong prompts include a clear task, some context about your situation, and the format you want back.
• Stay in the same chat and refine — “shorter,” “less formal,” “try again” is where the real quality comes from.
• Asking the AI to interview you, or to critique your draft as a specific skeptical reader, gets better results than asking it to just produce something.
• Draft freely, but verify any fact, number, or quote before you rely on it, and keep genuinely sensitive information out of the chat box.