For a long time, I thought I wasn’t good at using AI. Now you’d never know, considering I test and review AI for a living. But not so long ago, when you opened ChatGPT, Gemini, or Claude and typed in what felt like a well-defined request, you’d get a vague and unhelpful answer. Even with memory enabled, the responses were sometimes too long, too generic, and felt completely wrong. It’s not a hallucination per se, it’s just not good.
It occurred to me that the problem isn’t that chatbots are stupid. That’s what they’re guessing. After all, they don’t think like humans, and while some may understand context, they still give responses based on patterns.
This gap, between the user’s question and what the chatbot assumes the user meant, is the reason why most AI responses fail. ChatGPT or your favorite alternatives will fill in the blanks, make some guesses, and return responses that you’ll find helpful until you try it.
So I started using a single prompt to slow down the chatbot, clarify its goals, and force it to stop guessing.
I call this the “unicorn” prompt because it works with basically every AI tool I’ve ever used or tested. And once you start using it, it’s hard to go back.
“Unicorn” prompt
This is the exact prompt I use every Chatbot:
You pretend to be my assistant, but you actually want me to succeed. If you have any questions, please ask up to three questions. So please tell me the answer, the plan and the pitfalls. Keep it short and tailor it to: [insert goal]. If you need to make assumptions, list them first.
This prompt is perfect when you’re thinking, “I don’t know what to ask” or “I need something I can actually do.” do”Here’s a great real-life situation:
Here are 15 ways to use “unicorn prompts” with real-world examples.
- Write awkward text without sounding weird
goal: Write a text message to a friend you haven’t heard from in two weeks without making it awkward. - Send a friendly but firm email
goal: Please write an email to say that you are unable to take on any further work at this time, but that you would be happy to return next month. - React calmly to passive-aggressive messages
goal: Please respond politely to this message without being a doormat. [paste message] - Turn messy thoughts into clear sentences
goal: Please rewrite this in my voice in a clean, confident sentence. [paste draft] - Plan your week when you’re confused
goal: Plan your week with three priorities, five to-dos, and realistic blocks of time. i have [X] available time. - Create a to-do list that you can actually complete
goal: Turn that messy list into a realistic plan that includes things to do today and things to do later. [paste list] - Stop procrastinating and start now
goal: What are the first three small steps to start this task? [task] - Choose between the two options without thinking too much
goal: Please help me choose between option A and option B based on priority: [list options] - Plan your trip without opening 40 tabs
goal: Create a simple 2-day itinerary [city] Includes stops and breaks for children. - Make dinner plans when you’re tired and everyone is noisy.
goal: What are your top 5 easy dinner ideas using things you already have? [list ingredients] - Reduce your monthly expenses without sacrificing them
goal: Discover 5 practical ways to shave $200 off your monthly budget without making your life harder. - Write a “teacher message” that won’t make you feel anxious
goal: Write a message like the following to your child’s teacher [issue] That’s respectful but clear. - Learn something quickly without spiraling
goal: explain [topic] He’s smart but seems overwhelmed. Keep it short and practical. - Solve technical problems without guessing
goal: Troubleshooting why this happens [device/app] keep doing it [problem]. Please ask questions first. - Get better AI results for any chatbot
goal: Help us get the best answer from our chatbot for this task: [task]
Why it works so well
Most chatbots have the same bad habit of answering confidently, even if it’s wrong.
So if your prompt is missing important details, the AI will often automatically fill in the blanks. So you end up getting advice that may seem helpful but is completely wrong in certain situations.
Unicorn Prompt solves this problem in three ways:
- It forces clarification. Rather than guessing, chatbots ask questions first. That alone will improve the quality of your response more than any “magic words”.
- It forces structure. Don’t worry about walls of text. You will get clear answers to avoid mistakes.
- It keeps things short. This is a big deal. Most people don’t need a chatbot to write a novel. They need something they can skim and act on.
Take-out
This prompt works in almost all cases, but here are the moments where it saves you the most time:
- write a nasty message (text message, email, follow-up)
- plan your week When everything feels chaotic
- learn something quickly without spiraling into 10 tabs
- get stuck When your head is pounding and you just need a step
What I particularly like about this prompt is that it works well when switching between tools. ChatGPT, Gemini, and Claude each have different strengths, but this prompt makes them behave like the same type of assistant.
You can also add additional lines that will further improve the response.
“If you need to make a hypothesis, list it first.”
Try this universal prompt the next time you get stuck or switch from one chatbot to another. That’s the secret to getting a better answer almost every time.
to follow Tom’s Guide to Google News and Add us as a preferred source Get the latest news, analysis, and reviews in your feed.
