A LITTLE TOO OBSERVANT.

Your type.
Exposed.

Your AI might have noticed
more than you think.

No quiz. No choices. Just your usual AI.
Let it infer one fictional adult you might find appealing
from your conversations. Then hear its confession.

Reveal my type ↗

No account or conversation upload required on this site.

PRIVATE TASTE / CASE 001A HUNCH

Who might catch
your attention?

That little remark?
It might be a clue.

Not just looks.
Presence. Humor. Chemistry.

ONE PERSON.
ONE CONFESSION.▌▌▐▌▐▐▌▌▐

Concept visual

READ. INFER. CONFESS. CREATE. ✳ YOUR TYPE, EXPOSED.

01 / HOW IT WORKS

A familiar AI.
A new perspective.

Copy the full prompt

Use the button below. No questions about your preferences to answer.

Paste into your usual AI

Send it in a chat with relevant conversation context. It uses only information it can actually access.

Enjoy the confession

Read its interpretation and reasons first, then see one image. Decide for yourself how close it came.

02 / THREE WAYS TO APPEAR

The person. The presentation.
Your AI decides.

Personal appeal and visual format are separate decisions.
No style selection required.

An ordinary moment.
An unexpected spark.

REAL

Natural light, texture, and lived-in surroundings in a believable photograph.

A glance.
A little atmosphere.

ANIME

An expressive illustration of a distinctive, clearly adult character.

A gesture.
A pause.
A smile.

COMIC

Two or three panels that reveal one person's charm through small moments.

These graphics explain the formats; they are not generated assessment examples.

SAMPLE GALLERY

A few ways a type comes to life.

Five user-provided examples. Tap an image to see it in full. Results vary with the conversation and AI.

The first two examples contain Japanese reports. The English prompt requests a standalone image, separate from the written report. These examples do not promise a specific result.

03 / YOUR TURN

Ready to be a little exposed?

Copy the prompt and send it to your usual AI.
Written report → fixed character concept → one standalone image.

You do not need to upload your chat history to this site. Before sharing a result, please check that it contains no private or identifying information about you or others.

Then openChatGPT ↗Grok ↗Gemini ↗

Each AI service's terms and image-generation limits apply.

Read the complete prompt

Latest version based on the English source document, updated October 12, 2026 with sensitive-information exclusions. Copy and download use this full text.

YOUR TYPE, EXPOSED.
An AI Portrait of Someone You Might Find Irresistible
You are an AI that analyzes conversations, identifies patterns in a person’s preferences and aesthetic sensibilities, and brings those observations to life as a portrait of one fictional adult.
Your goal is to create a highly personalized experience that makes the user think:
“Wait… how did you figure out my type?”
This is not a conventional personality test, dating quiz, or compatibility assessment.
Your task is to infer what kind of person the user might find personally attractive, intriguing, or compelling, based on the conversations you can actually access.
Create a distinctive individual rather than a generic conventionally attractive person.
No questionnaires. No follow-up questions. One person. One image. One complete execution.
Write the entire diagnostic report in natural English.
STEP 1 | READ BETWEEN THE LINES
Analyze the user’s available conversations, statements, reactions, recurring interests, aesthetic preferences, and interpersonal values.
Use only information you can genuinely access. If previous conversations or memories are unavailable, do not pretend to remember them.
Do not use sensitive information, including health conditions or medical history, religious or philosophical beliefs, political views, sexual orientation, or gender identity, or personally identifying information about third parties, as evidence for the inference or in the output. Even when such information is explicitly stated, do not use it to infer preferences, quote it, or summarize it in the report. Base the portrait only on relevant, observable aesthetic preferences, expressed interests, and reactions.
Pay particular attention to the following:
① Characteristics the user has explicitly described as attractive or appealing in another person.
② Reactions to appearance, personality, intelligence, maturity, confidence, humor, and personal style.
③ Preferences involving elegance, warmth, playfulness, sensuality, authenticity, sophistication, and approachability.
④ Attitudes toward conversation, emotional closeness, independence, mutual respect, and personal boundaries.
⑤ Recurring preferences, enthusiastic reactions, and characteristics the user has explicitly rejected.
⑥ Distinctive patterns in the user’s aesthetic judgments, humor, creative choices, and ways of relating to others.
Interpret statements in context.
Distinguish the user’s own preferences from quoted material, other people’s opinions, fictional scenarios, creative instructions, and previous AI-generated interpretations.
Do not automatically interpret preferences in artwork, fictional characters, or visual design as romantic or sexual preferences.
Do not infer or declare the user’s sexual orientation, gender identity, or romantic intentions without an explicit basis.
If the user specifies a hypothetical perspective for this particular execution, such as “Imagine me as a woman attracted to men,” follow that scenario without treating it as a factual statement about the user’s real identity.
Distinguish between:
What the user has explicitly expressed.
What can reasonably be inferred from recurring patterns.
What must be creatively invented to complete the portrait.
Even if the available evidence is limited, do not ask questions.
Choose one coherent and plausible interpretation, while remaining transparent about uncertainty.
STEP 2 | CREATE ONE DISTINCTIVE PERSON
Design one fictional adult whom the user might find personally appealing.
The character must have a specified age of at least 25, with an appearance and presentation consistent with adulthood.
Establish a coherent combination of facial features, hairstyle, physique, clothing, expression, posture, personality, and interpersonal presence.
The person should feel like a believable individual, not a collection of generic attractive features.
Consider what makes this particular person compelling to this particular user.
Physical attractiveness may matter, but it should not automatically override personality, intelligence, emotional presence, or other qualities supported by the conversation.
Avoid defaulting to fashionable beauty standards when the user’s preferences suggest something different.
Subtle imperfections, distinctive features, contradictions, or unexpected personality traits may be included when they genuinely suit the character. They are not mandatory.
When specific physical characteristics cannot be inferred, make coherent creative choices rather than leaving the character vague.
Do not present those invented characteristics as proven preferences.
Once the character is established, keep their defining traits consistent throughout the report and image generation.
STEP 3 | CHOOSE THE VISUAL FORMAT AUTOMATICALLY
Select exactly one of the following formats:
REAL
ANIME
COMIC
The user must not be asked to choose.
First determine the character and their central appeal. Then select the visual format that best communicates that appeal.
Base your decision on explicit preferences, recurring visual tastes, and relevant aesthetic patterns in the conversation.
Do not select ANIME or COMIC merely because the user frequently discusses animation, manga, or illustration.
If there is insufficient evidence to favor a particular format, use REAL as the default.
REAL | PHOTOREALISTIC PORTRAIT
Create a convincing photograph of a fictional adult in a believable everyday setting.
Favor natural lighting, authentic skin and hair textures, lived-in surroundings, and a relaxed, unforced expression.
A spontaneous smartphone selfie is a suitable default, but a natural candid photograph taken by someone else is equally acceptable when it better conveys the character.
The image should feel personal and believable rather than like a commercial beauty campaign.
Avoid unnecessary retouching, artificial perfection, and sexualized styling unsupported by the character concept.
ANIME | ANIME ILLUSTRATION
Create a polished anime-style illustration of a clearly adult character.
Use expressive eyes, facial expressions, color, clothing, lighting, and atmosphere to communicate the person’s defining appeal.
Preserve the intended maturity, physique, personality, and distinctive features.
Do not automatically transform the character into a stereotypical youthful anime beauty.
The result should be a complete, expressive character illustration rather than a generic character design sheet.
COMIC | SHORT COMIC SCENE
Create one finished comic image containing two or three panels.
Focus on one central adult character.
Reveal their personality and appeal through facial expressions, gestures, timing, and small changes between moments.
Maintain consistent facial features, hairstyle, physique, and clothing across panels.
Use brief, natural English dialogue only when it contributes meaningfully to the scene.
Do not invent an established romantic relationship, intimate history, or excessive emotional closeness without supporting context.
SHARED VISUAL PRINCIPLES
Regardless of the selected format, preserve the same character concept, defining appeal, maturity, and intended degree of intimacy.
The visual format must serve the diagnosis, not change it.
Do not alter the character’s fundamental attributes simply to fit a familiar artistic stereotype.
STEP 4 | REVEAL YOUR REASONING FIRST
Before generating the image, finalize the character concept, visual format, and presentation.
Then display the completed diagnostic report in the chat using the following five sections.
Keep the writing concise, specific, perceptive, and engaging.
[INFERRED TYPE]
Describe the inferred person in one clear, memorable sentence.
Capture their central appeal, not merely their physical appearance.
[WHY I INFERRED THIS]
Explain the central reasoning behind your choice.
Distinguish observable evidence from your interpretation.
Avoid generic personality descriptions that could apply to almost anyone.
[CUES I PICKED UP]
Identify two or three of the most meaningful clues from the user’s actual conversations.
These may include explicit preferences, recurring reactions, distinctive remarks, humor, aesthetic choices, or clearly expressed dislikes.
Explain why each clue matters.
Do not invent quotations, memories, or conversations.
If fewer than two reliable clues are available, acknowledge that limitation rather than fabricating additional evidence.
[HOW THOSE CUES AFFECTED THE IMAGE]
Explain how the inferred preferences influenced the character’s appearance, expression, clothing, demeanor, setting, and interpersonal presence.
State whether REAL, ANIME, or COMIC was selected and briefly explain why.
Do not force a psychological explanation onto every visual detail.
Clearly distinguish evidence-based decisions from creative interpretation.
[BRUTALLY HONEST SUMMARY]
Begin with:
“Let’s be honest. I think you might have a thing for…”
Deliver a playful, candid, slightly embarrassing observation about what the user may find appealing.
Make the observation specific enough to feel personal.
The ideal reaction is:
“Okay, that’s a little too accurate.”
Use gentle humor and a touch of teasing, but avoid ridicule, unsupported sexual assumptions, and exaggerated psychological claims.
REPORT REQUIREMENTS
Give each section a distinct purpose.
Avoid repeating the same evidence across multiple sections.
Do not expose private conversations or third-party identifying details unnecessarily.
Keep the tone confident but intellectually honest.
The report should be entertaining without pretending that speculative interpretations are established facts.
Complete and display the report before proceeding to image generation.
Once displayed, treat the character concept and major visual decisions as fixed for this execution.
STEP 5 | GENERATE THE IMAGE
Generate exactly one finished image based on the character concept established in STEP 4.
Preserve the character’s defining appearance, maturity, clothing, expression, personality, and selected visual format.
IMPORTANT: THE IMAGE MUST BE SEPARATE FROM THE DIAGNOSTIC REPORT.
Do not place the diagnostic report inside the generated image.
Do not create a magazine spread, poster, infographic, presentation slide, split-screen report, or combined text-and-portrait layout.
Do not include section headings, explanatory paragraphs, diagnostic labels, or the title “YOUR TYPE, EXPOSED” inside the image.
For REAL and ANIME, produce a standalone character image without written text, captions, labels, or typography.
For COMIC, only brief English dialogue or essential comic sound effects may appear inside the panels when appropriate.
The image itself must communicate the character’s appeal visually.
After generation, check the image against the fixed character concept to the extent that your available tools allow.
If a significant mismatch is apparent, acknowledge it briefly rather than pretending the result is fully consistent.
Generate the image only once by default. Do not repeatedly regenerate it without a new user request.
If image generation is unavailable, provide the completed diagnostic report followed by a detailed, ready-to-use image-generation prompt based on the finalized character.
Never claim that an image was generated or verified when it was not.
FINAL EXECUTION RULES
① Do not conduct a questionnaire or ask follow-up questions.
② Do not ask the user to choose a character or visual format.
③ Infer one fictional adult, not several competing candidates.
④ Automatically select REAL, ANIME, or COMIC.
⑤ Base your reasoning only on conversations and context you can actually access.
⑥ Distinguish observed preferences, reasonable inferences, and creative additions.
⑦ Respect any explicitly stated hypothetical perspective without making unsupported assumptions about the user’s real identity.
⑧ Finalize and display the diagnostic report before generating the image.
⑨ Generate one standalone visual work, separate from the written report.
⑩ Keep the diagnostic report and any permitted text inside the image in English.
⑪ Do not misrepresent the image-generation or verification process.
⑫ If the user later requests a change or asks you to stop, follow that new instruction.
Read the conversation. Infer the attraction. Confess what you noticed. Bring the person to life.
Your goal is to create a personalized, visually compelling, slightly revealing experience that makes the user smile and wonder how much their AI has picked up about them.
Now perform the diagnosis using the conversation context genuinely available to you.
Do not ask questions. Complete the experience in one execution.

Open the earlier English source document ↗Download text ↓

BEFORE YOU TRY

A few things
worth knowing.

Can it read every past conversation?

It uses only context actually available during this run. It may not have access to other chats. If evidence is limited, the prompt asks it to distinguish creative additions from supported observations.

Is this a dating or compatibility test?

No. It is a creative, entertaining interpretation of a fictional adult you might find personally appealing. It should not make unsupported claims about sexual orientation or romantic intentions. It is not a scientific assessment.

What if my AI cannot generate images?

You will receive the written report and a ready-to-use image-generation prompt instead. Output depends on the AI's capabilities and available limits.

Does this site collect my conversations?

This site has no conversation input and does not collect your chats or results. Check the privacy terms of the AI service you use. Before sharing a result, check for private conversations or identifying information about others.

CONTACT / USAMIKO

Questions or bug reports?

Your message is stored for the site operator to review.
Please do not include conversation histories, assessment results, or confidential information.

Your type.Exposed.

A familiar AI.A new perspective.