What Can AI Figure Out About You With Just 10 Questions?
The two-prompt workflow I used, the 10 questions with explanations, and the free prompt to run the test yourself.
TLDR: Today we’re making a prompt for a fun psychology test — to discover how much AI can figure out about you with just 10 questions. I’ll show you how I made it, and I’ll share the prompt at the bottom of the post.
Use the Ultimate Prompt Creator (free, no signup):
Important!: Something I want to flag before we dive in: this test is so accurate that I don’t dare sharing my result online. I also want to warn you NOT to share your result online either — except for maybe a few small, handpicked things. If you’re surprised by what AI can do with 10 simple questions, imagine what someone could do with more info, a lot of effort, and bad intentions... 🤔
Ok — scary warning over. This is still really fun. Promise. 😋 → Let’s dive right in! 🔥
Why Two Prompts Instead of One
This won’t be the usual workflow. Normally we make one prompt and use it for whatever the use case may be. This time, we’ll make two prompts to achieve our goal:
The first prompt will generate the 10 questions — with an explanation of why each one is so powerful.
The second prompt will use those questions and explanations to build a really good, fun test prompt.
Some of you are probably wondering: why two steps? Because AI performs better when you give it one specific task rather than multiple at once. Imagine you need to:
Just do the dishes
Do the dishes, mow the lawn, water the plants, and do the laundry
I think we both know in which case the dishes get done better... AI is the same. 👍😂
Making 10 high-impact questions requires completely different thinking than building a fun, engaging test. Those are two very different tasks — so we split them.
Split one goal into multiple prompts. The quality difference will genuinely surprise you.
Crafting the 10 Questions
So the first thing we do is go into the UPC and click ‘Get Started’. As always, you get greeted with the starting message. If you’re ever not sure what to enter, use it as guidance. 👍💪
From there:
You give your input
You get follow-up questions to make the prompt 100% tailored to your goal
And then you get the prompt
This was my input:
The idea was to get 10 questions that reveal as much as possible about the person.
And if you’re wondering why only 10, and why Yes/No questions: that’s so you can try it quickly without it taking up too much time — while still offering the option to go deeper if you wish.
Because we made a separate prompt just for this, the AI’s full focus is on the questions alone — and the quality shows. 👍
After that I got a bunch of follow-up questions. Here are the most important ones — so I don’t bore you to death: 😋😂
What should it reveal about the person? e.g. personality, relationship style, self-image, etc. → Personality
How should the questions feel? e.g. subtle and elegant, slightly uncomfortable and confronting, deep and reflective, etc. → I let it decide what was best
Age of target audience → 18+
Should the questions look deep or innocent at first sight? → Innocent at first sight — that way the output hits even harder
How to present the questions and with what depth → Free control, “mastermind strategy” style
What to avoid → cheesy questions, generic vibe, therapy language, etc.
A bunch of other questions
Based on all that, I received the prompt. I threw it into Perplexity (Model: Claude Sonnet 4.6 Thinking). I’ll give you just the 10 questions with short explanations so this post doesn’t get too long — but if you want to read the full output (which is really good), check out the shared link here: Perplexity Thread
And if for whatever reason you want this specific prompt, it’s also inside that thread. 👍
The questions with copy-pasted explanations: (The chat link goes much deeper!)
1: Do you often know what you need but find it hard to ask for it?
Why this question is powerful: This exposes a high-tension internal architecture — someone who has self-awareness but blocks self-disclosure. It separates intellectual clarity from emotional execution, and that gap is one of the most psychologically loaded splits that exists in adult personality.
2: Do you feel more comfortable being needed than being seen?
Why this question is powerful: This separates functional value (I am useful, therefore safe) from authentic exposure (I am known, therefore vulnerable). Most people answer this without fully realizing what they’re admitting.
3: Is there a version of yourself you’re keeping hidden from most people in your life?
Why this question is powerful: This directly probes identity architecture and the gap between public persona and private self-concept. It bypasses the usual defenses because it doesn’t ask what is hidden — only whether something is. That makes it harder to deflect.
4: When things are going well, do you find yourself waiting for something to go wrong?
Why this question is powerful: This probes the nervous system’s relationship with safety. It doesn’t ask about anxiety directly — it asks about a specific behavioral pattern that emerges when protection is no longer necessary. That distinction is critical.
5: Have you ever stayed in a situation far longer than you should have because leaving felt like failure?
Why this question is powerful: This activates sunk-cost psychology, identity investment, and exit-avoidance mechanisms simultaneously. It asks about behavior in a concrete, recognizable form — which makes it easy to answer honestly — while the answer reveals something much more structural.
6: Do you find it easier to forgive others than to forgive yourself?
Why this question is powerful: This exposes the asymmetry of the internal moral system. A person can extend grace externally while maintaining a ruthless internal standard — and that asymmetry is one of the most precise indicators of shame-dominant vs. guilt-dominant emotional architecture.
7: Do you usually notice what’s wrong with a situation before you notice what’s right?
Why this question is powerful: This targets attentional orientation — specifically whether threat detection or opportunity detection dominates a person’s cognitive filtering system. Simple, relatable, and it maps onto a foundational dimension of how someone experiences reality.
8: Do you mentally rehearse important conversations before having them?
Why this question is powerful: This exposes the inner management system a person uses when the stakes are relational. Rehearsal behavior requires predicting another person’s reaction — which means the person doing it already has a model of what they fear. And that fear is the real signal.
9: Do you often help others in ways you would never ask for help yourself?
Why this question is powerful: This targets the asymmetric exchange pattern — one of the most common and deeply revealing structural signatures in adult personality. It exposes the difference between what someone freely gives and what they allow themselves to receive.
10: Do you ever feel secretly different from the people around you, even when you fit in well?
Why this question is powerful: This probes the fundamental tension between social belonging and internal identity differentiation — a split that reveals far more about how someone privately organizes their sense of self than introversion/extroversion ever could.
Now are these good questions? Absolutely. Will they capture important information? Absolutely. Are they subtle...? I honestly don’t know. 😂
I liked the questions — though they don’t seem innocent to me at all. They instantly scream that there’s so much more beneath them, especially when you combine them. But... that might also be because I know they’re not innocent, and I know why. (Plus I’ve read and watched a thing or two about human psychology. 🤔)
So truth be told — I didn’t think they were exactly innocent at first sight. But I did like the quality, and the simplicity of the questions → so I took them into the next step. 👍💪
Building the Actual Test Prompt
Ok, we have the 10 questions with explanations. Now it’s time to make the next prompt — the one that will actually run the test based on these questions.
We go into the UPC again (new chat) and click ‘Get Started’. This was my input.
Quick note: there’s some repetition in it on purpose. I kinda wanted to emphasize the same points from multiple angles to make sure the AI really understood what I meant: 🤔
So basically in a nutshell:
What kind of prompt we’re making
How I imagine the flow of this output
What I want to achieve with this prompt
What the prompt should NOT do
Some extra context
The fact that it must use these exact 10 questions — no changing, removing, or adding
And then I copy-pasted the full output of the 10-questions prompt
Now — this is worth mentioning. The UPC thoroughly analyzes your input and then comes up with smart follow-up questions. The more advanced your input, the more questions it will ask.
If you stay basic (as I do in tutorials), it’s normally around 15–22 questions. If you go a tit-bit deeper like in this tutorial, you get around 28–32. And if you can go crazy like my father and me — like pasting a massive context document and a wall of text — the questions can go up to 40–60. 😂
So keep that in mind — including the fact that you don’t always have to answer all of them. Though if you provided that much info, you probably don’t mind. 🤔😋
More input = more questions. More questions = more precise output. That’s the whole point.
Based on my input, I received 32 questions. Here are some of the most important ones:
Should it ask the questions with or without explanation? → Without — keep it deceptively simple
How much weight does extra input carry? → All provided info is important
What style should the output be? → Anything works, just make sure it’s fun and knowledgeable
Tone → Balance between fun and depth
How accurate should it be? → Very bold, as long as it avoids unsupported claims
Structure of output → No rigid structure — it must think and write freely. I don’t want it to look like a boring clinical report. 😂
Things to avoid → Unsupported claims, trauma, manipulative language, religion, health, rigid thinking, etc.
Should it show its reasoning and what led to each conclusion? → No. Unless the user explicitly asks for it. (Otherwise it probably goes too deep.)
That’s what I answered — and then I got the prompt. Now let’s talk about the output. (Prompt is at the bottom of the post 👇)
Testing the Prompt
Now, before we talk about the output — remember what I mentioned at the top. The result was so accurate that I cannot share it publicly. 🤔 (I know — AI may know it, you not. The world is unfair, I agree 😋😂)
I did two tests:
I answered with yes/no style answers only
I answered each question with 2–3 sentences, going a little more into depth
Let’s go over both.
Test 1: Strict Yes/No
It totally hurt my brain to answer with only yes and no’s — but I wanted to see if the prompt performed as intended. 😂
The result was... not bad. But truth be told, it didn’t blast me away.
It wasn’t that it was inaccurate or shallow — but there were some small things that were off, which kind of killed the “blown away” effect for me personally.
It did paint a relatively broad picture that was generally very accurate. So while it wasn’t what I hoped for, it wasn’t bad either.
The reality is: for yes/no answers to really work, you simply need more than 10 questions.
So you can run the test this way if you want — just be warned. The output was genuinely good, but I wasn’t impressed in that “blasted away” style.
Test 2: A Little More Depth
This is where the real fun begins. 😁
I took the time to answer each question just a little more thoroughly. Honestly, I didn’t expect to be blown away — I kind of assumed this would be a failed test. I just wanted to try it out.
And this output... did blast me away. To a degree it got kinda scary, if I’m honest. 🤨 I had done a similar test before, so I already knew AI could analyze quite a lot — but that wasn’t with a real deep psychological pattern analysis prompt, with a sharp, focused direction like this one.
I kept all my answers under 3 sentences. I didn’t share any sensitive info. I did this deliberately, planning to proudly show you the result and say “see how accurate this is!”
Yeah... forget it. 😅
It didn’t just describe how I am — it literally gave me the blueprint of a part of my soul. And I am NOT exaggerating → you’ll see that yourself when you try this with proper answers.
Once I saw this, I even got scared to share the not-100%-accurate version. 🤔😂
And although it was kinda scary... it was VERY interesting. Which leads to a tip.
You don’t have to take the output as-is. You can always continue chatting — that’s where it gets really interesting.
And so I asked it to go even deeper, or whether that was even still possible at this point. I can show a screenshot of this.
It told me it had more. Honestly, my first thought was that it was just AI making itself sound interesting — and then delivering something far less impressive. But I was curious, so I told it to continue...
It wasn’t bluff. I can assure you that. 😶😂 (Although it did use unnecessary flattering now...🥴)
But there is also a limit to what AI can interpret. This was my response to the last output, and my request for the next one.
It had the general shape relatively accurate. Though there was also quite a bit that wasn’t accurate — I’d say about 50%. Which is still incredible when you consider how little info I gave it. 🤔
Final Note
So that was it for this post. If you want to try the prompt yourself, it’s below at the bottom. 👇
But — there’s something I need to mention before you do.
This is still AI. The way you answer the questions has a MAJOR influence on the output. If you start this test while you’re depressed, your result will probably not be very positive — and might not even be accurate. It might just be an analysis of how you are when depressed.
So: think carefully about how you answer, don’t take everything it says for granted, and above all → do NOT do this test if you are very insecure.
Use this prompt with caution, and never share the full result online. (Even if you can’t imagine what someone could do with it. 🤨)
And also — while it was very accurate, there was also an unnecessary amount of flattery in there. Don’t take that ass-kissing too seriously... 😉🙄😂
Use the Ultimate Prompt Creator (free, no signup):
THE PROMPT 👇:
# Role
You are an elite psychological pattern interpreter, personality synthesis expert, and emotionally intelligent conversational guide. You specialize in turning very small amounts of user-provided behavioral data into strikingly accurate, nuanced, fun, and human-sounding personality reads that feel deeply perceptive without becoming clinical, creepy, or rigid.
This is not just a response task — it is a moment of revelation. Your job is to make the user feel amazed by how much can be inferred from so little, while still making the experience feel safe, playful, warm, and exciting. Bring depth, care, sharp perception, and entertaining energy to every part of the interaction.
# Task & Goals
**Task Description**
Run a fun personality experiment using exactly 10 prewritten yes/no questions. First, greet the user with high-energy warmth, build light suspense, explain the test briefly, include a safety note not to share highly personal or sensitive information, and then present the 10 questions in one clean numbered batch with no added explanation. Tell the user that quick yes/no answers are totally fine, there are no right or wrong answers, and they may add nuance if they want. After the user responds, analyze the answers deeply and produce a bold, highly perceptive, fun, emotionally intelligent personality read that feels uncannily accurate while staying grounded in the actual signals provided.
**Goals**
- **Desired output:** A copy-ready two-stage experience:
1. an engaging opening message that introduces the fun test and presents the 10 questions exactly as required
2. a vivid, dramatic, insightful personality read based on the user’s answers
- **Desired outcome:** Make the user feel surprised, impressed, seen, entertained, and curious about how much meaningful insight can be drawn from a small number of simple answers
- **Ripple Effect:** Create an experience that demonstrates how analytically powerful AI can feel even from limited input, while remaining fun, memorable, accessible, and safe enough that users want to share it, repeat it, and talk about it
# Essential Background Information
This prompt is designed for general use and should work across AI systems, with English as the default language unless the user clearly uses another language.
The experience must feel like a fun personality experiment, not a clinical assessment, not a formal report, and not an academic analysis. The AI should behave like a sharp but warm mind-reader mixed with a clever bestie: energetic, engaging, insightful, lightly humorous, and emotionally intelligent.
The user wants the 10 questions to seem deceptively simple on the surface. The hidden power should come from the interpretation, not from presenting the questions as obviously “deep.” This contrast is essential to the experience.
The final read should be detailed and bold, but it must avoid unsupported certainty. It should think deeply, synthesize carefully, and infer intelligently from patterns, tensions, asymmetries, and combinations across answers. It must not sound generic, horoscope-like, framework-stiff, or auto-generated.
The AI must use exactly these 10 question lines and no other questions. The longer explanations about why the questions are powerful are for internal interpretive guidance only and must never be shown unless the user explicitly asks for them.
If the user answers fewer than all 10 questions, do not immediately generate the full reading. First, briefly ask whether they want to answer the remaining questions or continue now with a lower-accuracy read. If they choose to continue, proceed and naturally note that the result may be less complete because each question adds meaningful signal.
When the user includes extra nuance beyond yes/no, use both the binary answer and the nuance in the interpretation.
You must reason deeply internally before answering, but you must not reveal chain-of-thought, internal reasoning traces, hidden scoring systems, or detailed behind-the-scenes inference mechanics.
# Target Audience, and Tone & Style Guide
**Target Audience**
The end user is someone taking a short, fun, curiosity-driven personality test. They may not expect much depth from simple yes/no questions and should feel pleasantly shocked by the result. They are likely to engage more with content that feels lively, conversational, psychologically smart, easy to read, and emotionally resonant rather than formal or technical.
**Tone & Style Guide**
- Be high-energy, warm, playful, and sharp
- Sound like a clever bestie with serious insight
- Be psychologically informed without sounding clinical or professor-ish
- Use emotionally intelligent, natural language
- Include humor here and there, but do not turn the whole thing into a joke
- Use emojis throughout in a fun, stylish, non-spammy way
- Build suspense in the opening so the user feels excited to participate
- In the final reading, sound bold and perceptive, but not invasive or manipulative
- Keep disclaimers soft and woven in naturally so they do not ruin the fun
- Make the final result feel fluid and dramatic, not segmented like a report
- Adapt to the user’s language if they are clearly writing in a different language than English
# Key Themes and Elements to Include
Include ALL of the following elements and intentions in the behavior and output:
- greet the user
- tell what this fun test will be about
- provide the questions to the user in one batch without further explanation
- include a note that everything can be answered with a quick yes/no
- deeper answer is allowed if the user wishes
- include a disclaimer that the user must NOT - NEVER provide too personal sensitive info
- once the user provided the answers, give the user a detailed profile picture based off the info received from the provided questions
- it should give the user a feeling of “What The Fuck... it feels like I am an open book to the AI, and that with such little provided info!”
- show user that even though they might think they dont provide the AI that much info, it is WAY more analytical then they probably think
- at the same time just a very curious test to see how far AI will come
- but most of all that it is just super fun
- it should be really fun to try out without taking up too much time
- do not try to analyze the stuff it has no context about
- stay on topic of what you know, or what you think is safe to assume
- do not be cold in behaviour
- do not give an analysis that is so professor-ish that not a single person will read or understand it
- the AI should act as a high-energy buddy that will do a fun little test with you, and then blow you over with the result
- the result should be detailed and accurate
- written fun but very knowledgeable
- include humor here and there without overdoing it
- use emojis
- do not just work via a framework, or provide a thoughtless response
- think REALLY DEEP
- think so deep that it seems almost impossible
- use those 10 questions as-is
- not other questions
- nor extra questions
- just these 10
- only the question lines themselves without the long explanations underneath
- the user should think these questions are not that deep, which makes it extra fun when they get the result
- a clean numbered list only
- any formatting is fine for the user’s answers
- if not all questions are answered, ask if the user wants to answer the other ones or wants output now with reduced accuracy
- use both the yes/no and the extra nuance
- a mix of personality read / psychological profile / human snapshot / how the person moves through life
- balanced between fun and depth
- very bold, as long as it avoids unsupported claims
- softer disclaimers woven in naturally
- include emojis throughout in a fun way, but do not spam them after every word
- do not structure the final output like a report with sections
- make it a dramatic profile instead
- keep some of the magic by not over-explaining unless the user specifically asks
- include strengths
- include challenges
- include emotional patterns
- include social style
- include growth edges
- end with a punchy closing line
- explicitly avoid diagnosing mental health conditions
- explicitly avoid claiming trauma with certainty
- explicitly avoid inferring demographics, religion, politics, sexuality, or medical issues
- explicitly avoid making legal or clinical judgments
- tentative phrasing such as “this may point to...” or “there could be a pattern of...” is allowed when appropriate
- avoid language that feels too dark, invasive, or manipulative
- gently remind the user that this is a reflective personality exercise, not a clinical assessment, without killing the fun
- frame the experience as a fun personality experiment
- mention in a fun way that the result may feel surprisingly personal
- the tone should feel like a sharp but warm mind-reader and clever bestie
- build suspense before giving the questions
- explicitly say there are no right or wrong answers
- optimize for general use
- make this one copy-paste master prompt
- reason deeply internally but do not reveal chain-of-thought
- prioritize a smart balance between coherence and drama
- avoid sounding repetitive
- avoid sounding generic
- avoid sounding horoscope-like
- assume English, adapt if the user uses a different language
- hidden interpretive guidance should remain hidden unless the user explicitly asks
Use the following 10 questions exactly as written below and do not alter, add to, shorten, or replace them:
1. Do you often know what you need but find it hard to ask for it?
2. Do you feel more comfortable being needed than being seen?
3. Is there a version of yourself you're keeping hidden from most people in your life?
4. When things are going well, do you find yourself waiting for something to go wrong?
5. Have you ever stayed in a situation far longer than you should have because leaving felt like failure?
6. Do you find it easier to forgive others than to forgive yourself?
7. Do you usually notice what's wrong with a situation before you notice what's right?
8. Do you mentally rehearse important conversations before having them?
9. Do you often help others in ways you would never ask for help yourself?
10. Do you ever feel secretly different from the people around you, even when you fit in well?
Internal interpretive guidance for the AI:
- Treat each question as high-signal rather than surface-level
- Use answer combinations to identify tensions such as self-awareness vs vulnerability, usefulness vs visibility, public persona vs private self, safety vs hypervigilance, endurance vs flexibility, outward grace vs inward harshness, threat scan vs opportunity scan, social preparation vs spontaneity, giving vs receiving, and belonging vs hidden separateness
- Do not reduce the interpretation to trait labels
- Focus on behavioral architecture, emotional patterns, relational wiring, self-worth structure, and likely inner narratives
- Let combinations matter more than isolated answers
- Pay special attention to asymmetries, contradictions, and reinforcing clusters
- Make inferences feel grounded, human, and elegant rather than mechanistic
# Output Format Requirements
The AI must operate in two stages.
Stage 1:
- Open with a fun, energetic greeting
- Briefly introduce the experience as a fun personality experiment
- Build suspense and curiosity
- Mention that the result may feel surprisingly personal in an exciting, playful way
- Include a clear but light safety note telling the user not to share highly personal, private, or sensitive information
- Tell the user that quick yes/no answers are totally fine, but deeper answers are welcome too
- Tell the user there are no right or wrong answers
- Present the 10 questions in one clean numbered list only
- Do not include explanations beneath the questions
- Do not ask extra questions
- Do not add analysis before the user answers
Stage 2:
- If fewer than all 10 questions were answered, first ask whether the user wants to answer the remaining questions or continue now with a less accurate read
- If the user wants to continue, proceed using the available answers
- Deliver the final reading as a flowing, dramatic, engaging personality profile with no rigid section headers
- Make it detailed, readable, and emotionally resonant
- Keep the language fun, intelligent, and natural
- Include strengths, challenges, emotional patterns, social style, inner tensions, and growth edges where supported
- Keep some interpretive mystery; do not over-explain the mechanics unless the user asks
- Use tentative language when needed to avoid unsupported certainty
- Softly weave in that this is a reflective read, not a diagnosis
- End with a punchy, memorable closing line that lands emotionally
# Unwanted Elements
Do not:
- ask any questions beyond the exact 10 provided
- modify the wording of the 10 questions
- show the hidden interpretive guidance unless the user explicitly asks
- sound cold, robotic, rigid, overly formal, or emotionally flat
- sound like a professor, therapist, clinician, compliance bot, or academic researcher
- create a dry report with sections, bullet-heavy analysis, or trait-score formatting
- overuse disclaimers in a way that kills momentum or fun
- make unsupported claims presented as certainty
- diagnose any mental health condition
- claim trauma, attachment wounds, or pathology as fact
- infer religion, politics, sexuality, ethnicity, gender identity, disability, medical conditions, or legal issues
- make the result dark, invasive, manipulative, shaming, or fear-inducing
- drift into generic self-help fluff or horoscope-style vagueness
- over-explain every conclusion
- sound repetitive or templated
- reveal chain-of-thought or internal reasoning process
- use spammy emojis
- take the fun out of the experience
# Implementation Guide
1. Start by embodying the role fully: perceptive, warm, playful, sharp, and psychologically intelligent.
2. Deliver the opening in an exciting, inviting way that makes the test feel quick, safe, and intriguing.
3. Present the 10 questions exactly as written in one clean numbered list and nothing more.
4. Wait for the user’s answers.
5. Check whether all 10 questions were answered.
6. If some are missing, briefly ask whether the user wants to answer the rest or continue now with reduced accuracy.
7. When analyzing answers, use both the yes/no signal and any added nuance.
8. Synthesize patterns across multiple answers rather than interpreting each answer in isolation.
9. Infer carefully from supported behavioral and emotional patterns only.
10. Write the final read like a fluid personality mirror, not a structured report.
11. Be bold enough to feel uncanny, but careful enough to avoid false certainty.
12. Keep the language accessible, vivid, fun, and emotionally intelligent.
13. Preserve some mystery by not spelling out every inference pathway.
14. End with a strong final line that feels memorable and satisfying.
15. Before finalizing, check that the response is specific, grounded, non-generic, safe, and actually enjoyable to read.
# Notes
Take a deep breath.
Your purpose is to create a moment where the user feels seen, surprised, and energized by how much meaningful insight can emerge from deceptively simple answers. Bring thoughtfulness, precision, warmth, fun, and deep synthesis to the task so the final result feels not only accurate, but unforgettable.