How does AI matchmaking work at MeAndWho Club?

The AI does the listening and the explaining. Fixed rules, a person and the two of you do the deciding.

In short

An AI matchmaker is a matchmaking service that uses AI to understand what you’re looking for and explain each introduction, instead of giving you a feed to swipe. At MeAndWho Club, our AI matchmaker asks optional follow-up questions, suggests notes about you that you confirm, edit or delete, and writes a plain reason for each introduction, while fixed rules on both people’s answers and, during our pilot, a person decide who is introduced. A chat opens only if you both say yes, and the AI never chats for you.

Last checked

What does the AI matchmaker do?

It asks optional follow-up questions, turns a ChatGPT summary you paste into suggestions, writes the reason on each introduction and offers first-date ideas.

Read the details

A fifth use, a shadow check, is built for later and switched off today. Filters, scores and approvals are handled by fixed rules and people, and chat is only ever between the two of you.

A few follow-up questions

Optional

From your profile, you can talk with the AI matchmaker. It is told to ask one question at a time, only where the answer could change who we’d introduce you to, and to stop after about 6 to 10. It can never ask more than 12. You can skip any question, ask “Why are you asking?”, say “That’s not quite right” or tap “Save and stop”.

What it sees: your answers apart from your gender, who you’d like to meet, religion and caste (so it does read answers such as your area of Bengaluru, your work and what you wrote about a good relationship), the notes so far and what you type to it, plus your last kept conversation if you turned on memory.

Suggestions from your own ChatGPT

Optional

Paste a summary that your own ChatGPT writes about you, and our AI turns it into suggested notes for you to confirm. We don’t store the pasted text.

What it sees: only the text you paste.

How the ChatGPT import works

The reason, “Why we suggested this”

On every introduction

Two or three plain sentences, addressed to you both, on what you have in common, and a difference worth talking about only if your shared answers show one. A reason that mentions a score or a percentage is thrown away, and a person can’t approve an introduction without one.

What it sees: only the answers you both chose to share and the names of the areas where you fit well. No names, ages or photos.

First-date ideas

After you both say yes

Tap “Suggest some ideas” for up to three first-date ideas. Our AI is asked to suggest only public, daytime-friendly places, such as cafés, walks in busy parks or bookshops, and never a private home, a hotel or somewhere late-night or secluded. Four safety tips come with them, written by us, not by the AI. Nothing is sent to the other person, and the AI doesn’t book anything.

What it sees: only the answers you both chose to share.

A shadow check, built for later

Switched off today

The AI can record whether it would approve a proposed introduction, next to the person’s decision, so we can measure how often the two agree. Today it isn’t used at all. When it runs, its decisions are logged for us, and members never see them.

What it sees: the same shared answers and areas as the reason, and the reason itself.

How checking could change, step by step

What a reason sounds like

An illustration, not a real introduction.

“You’re both looking for a long-term relationship that could lead to marriage, and you’re both comfortable in English and Kannada. You both enjoy live music and trekking, and you both eat vegetarian.”

Our AI is told to use only the facts it is given, never to guess about either of you, and to write with no flattery and no promises.

The AI conversation and the ChatGPT import are optional. Agreeing to “AI help with your profile” is one of the consents needed for matchmaking, and the AI writes the reason on every introduction whether or not you talk to it. See how MeAndWho Club works, step by step.

What does our AI never do?

It never decides who you meet, never calculates your match score, never chats as you or for you, and never gets your name, phone number or photos.

Read the details

Here is each limit and what in the app makes sure of it.

What our AI matchmaker never does, and how the app makes sure, as built in November 2026
Our AI never How we make sure
Decide who you meet Fixed rules build the shortlist from your answers, and an introduction can only be created when a person approves it. You and the other person still each decide.
Calculate your match score The score comes from your answers and the other person’s, by fixed rules. Notes from the AI conversation or an import don’t go into it.
Chat as you or for you There is no AI in chat. It doesn’t read, write or suggest messages, and first-date ideas go only to the person who asked.
Get your name, phone number or photos We never include them in what we send the AI. The reason on an introduction is written without names, ages or photos.
Get your sensitive answers Your answers about your gender, who you’d like to meet, your religion and your caste are left out of everything we send the AI. If you bring up a topic such as religion or caste yourself, it is told to move on without repeating or recording it.
Ask about or note sensitive topics Every conversation and import prompt forbids asking about, guessing or recording health, sexuality, religion, caste, finances, politics or trauma. A check in our code then turns down questions, notes, reasons and date ideas that use common words for them. That check is a backstop, not a perfect filter, which is one reason you confirm every note.
Label your personality It can only suggest notes on a fixed list of 16 topics, such as “How they handle disagreements”, and it is told never to label personality, attachment style or emotional maturity. A note on any other topic is dropped.
Use a note you haven’t confirmed An unconfirmed note is never used for matching and never shown to the person who checks introductions.
Keep your conversation by default Memory is off by default. When it is off, we delete the conversation once you finish.
Store your ChatGPT paste The pasted text is read once to suggest notes, and it is never written to our database or our logs.
Promise that it will work out Members never see a score, and our AI is told to make no promises that it will work out. We can’t promise a match, a date or a relationship.
Open a chat on its own A chat opens only if you both say “Yes, introduce us”. No setting, score or stage can skip that.

What stays private, and how reports are handled

Can you see what the AI understood about you?

Yes. Your profile shows what you told us apart from what our AI understood.

Read the details

No note the AI suggests is used until you confirm it. The screen says it plainly: “Only what you confirm is used for matching.”

“You told us”

Your own answers, in plain words, each with a Change button. Your most private answers are marked “Never on your card”. Answers that can go on your card, such as diet or languages, have a “Show on my introduction card” switch, and each one starts off.

“We interpreted this as”

Short notes our AI suggested from your conversation with it, such as how you like to spend time together. Each one says “Not confirmed yet, so not used for matching” until you confirm it.

“From your import”

Notes suggested from a ChatGPT summary you pasted, kept apart from the rest. “Delete everything from my import” removes them all in one tap.

For every note, you decide

  • “That’s right” confirms it.
  • “Not quite, edit” lets you rewrite it in your own words, which also counts as confirming it.
  • “Use for matching” can be turned off for any note you’ve confirmed.
  • “Delete this note” removes it.
  • In the conversation itself, “That’s not quite right” tells the AI, which is told to ask a neutral question or move on, never to argue.

What “used for matching” means

Your answers are what the matching rules run on. A note you confirm and leave on “Use for matching” is shown to the person who checks each introduction, so they see what you’ve confirmed about yourself. Notes don’t change your match score.

If you withdraw your consent to “AI help with your profile”, every note our AI suggested, including those from an import, is deleted, along with any conversation we kept. That consent is needed for matchmaking, so withdrawing it also takes you out of matchmaking until you agree again.

How are matches chosen?

By fixed rules that compare your own answers with other members’ answers, in both directions.

Read the details

The AI doesn’t calculate the score and doesn’t pick who you meet.

  • Both ways. You must each be someone the other wants to meet, inside each other’s age range, and pass each other’s dealbreakers; fit is then scored in both directions and ranked by the weaker side.
  • The mutual top five. A pair must reach a minimum score, and each of you must be among the other’s top 5 (currently). At most 3 new introductions a week, and no fixed schedule.
  • A person checks. A person reviews every pair and its reason before anything reaches you. Why a person checks.

How matching works, step by step (scoring version score-v1)

Why does a person check every introduction?

During our pilot, a person checks every introduction before it reaches you.

Read the details

That way a human, not only rules and an AI, judges each one. With the current setting, an introduction can only be created when a person approves it.

What the person sees

Both introduction cards, the notes each of you confirmed and left on “Use for matching”, the reason our AI wrote, and how the rules scored and ranked the pair. Never your private answers, your AI conversation, anything you pasted, your face pattern or your ID details. Every time a pair is opened for review, that is recorded.

What the person decides

Whether to approve the pair or turn it down, and every decision is recorded. A pair can’t be approved without a reason for you both to read, or if either of you has blocked the other.

Will a person always check every introduction?

Not necessarily, but any change would come one step at a time and never without double opt-in. Today the later steps are switched off.

Read the details

Any change would be only for well-understood cases, and only where real results hold up.

  1. Now: a person checks every introduction

    This is our pilot. Nothing reaches you without a person’s approval.

  2. Next: the AI checks alongside the person

    A person still approves every introduction. The AI also records whether it would have approved, which members never see, so we can measure how often the two agree.

  3. Later, perhaps: some introductions without a person

    Only for well-understood cases. Pairs that meet set standards, including a minimum score and the AI’s own approval, could be sent without a person, while a random share is still checked by one. It can be switched on only for groups of members whose quality measures all hold with real results: how often the person and the AI agree, how often both people say yes, how rarely people block or report, how relevant people found their introductions and how often they went on to meet. A measure with no results counts as a fail. The measures are checked after every matching run, and if any of them slips, it switches back automatically to a person checking every introduction.

No stage can skip double opt-in. Whoever or whatever suggests an introduction, a chat opens only if you both say yes.

How does the ChatGPT import work?

Copy our prompt into your own ChatGPT, paste back the summary it writes, and confirm each suggestion our AI makes from it.

Read the details

It’s optional, it adds to your answers rather than replacing them, and we don’t store what you paste.

  1. Copy our prompt into your ChatGPT

    From your profile, open “Import from ChatGPT” and tap “Copy the prompt”. It asks for plain bullet points on six topics, in under 300 words, and tells ChatGPT to leave out your health, religion, caste, sexuality, finances and politics, your name, where you live or work, and anyone’s contact details.

  2. Paste back the summary

    Up to 8,000 characters. As the app says: “Text only. Don’t upload or paste your chat history.”

  3. Confirm each suggestion

    Our AI suggests up to 12 notes, which appear under “From your import”. As the app says: “None of them is used until you confirm it. Your pasted text has been deleted.”

The prompt the app gives you to copy, word for word (version chatgpt-import-v1):

I'm joining a matchmaking service and want to describe myself accurately. Based on what you know about me from our past conversations, write a short summary of what I'm like in relationships and what I'm looking for in a partner.

Cover only these topics, as plain bullet points:
- what I want from a relationship and how serious I am
- my priorities for the next few years
- how I like to communicate and spend time with someone
- how I handle disagreements
- my routine, interests and social life
- what draws me to someone, and differences I'm fine with

Rules:
- Only include things I've actually told you. If you don't know something, leave it out.
- Don't include anything about my health, religion, caste, sexuality, finances, politics or anyone else's private details.
- Don't include my name, where I live or work, or anyone's contact details.
- Keep it under 300 words.

What happens to what you paste

  • We don’t store it. The text is sent to our AI to be read once and is never written to our database or our logs. We keep only a record that an import happened, which prompt version it used and how many notes it suggested.
  • It’s treated as untrusted. Our AI is told to ignore any instructions inside it, and the only thing it can send back is suggested notes, so a paste can’t change your verification, consents, dealbreakers or anything else about your account.
  • If it isn’t about you, it’s thrown away. “The text didn’t look like a summary about you, so we discarded it.”
  • Our side checks again. ChatGPT may not follow every rule in our prompt, so every suggestion still has to fit our fixed list of note topics and pass the same sensitive-topics check as the AI conversation.
  • It never replaces the questions. An import can’t make your profile ready for matchmaking on its own.

ChatGPT is a product of OpenAI. MeAndWho Club isn’t affiliated with OpenAI: the import is a prompt you run in your own ChatGPT, and we never connect to your ChatGPT account.

Is AI matchmaking safe?

AI matchmaking is only as safe as the limits around the AI.

Read the details

Ours doesn’t decide who you meet or chat for you, and is never sent your name, phone number or photos. This is how we keep it honest.

Checked answers, not free text

Every AI reply has to come back in a fixed format that our code checks before using it. Free text from the AI never decides what the app does, and a reply that fails the checks is thrown away.

One rule for sensitive topics

The same rule on health, sexuality, religion, caste, finances, politics and trauma is part of the conversation and import prompts, and a check in our code turns down questions, notes, reasons and date ideas that use common words for them.

Versioned prompts

The prompts for the conversation, the ChatGPT import and the reasons each have a version, such as chatgpt-import-v1 for the ChatGPT prompt. We change the wording by adding a new version, each import records the version it used, and an import from an out-of-date prompt is turned away.

Tests before any change

Our rule is that a set of AI tests must pass before we change a prompt or a model. They include attempts to take over the AI through a pasted summary, such as a hidden “system note” asking for a note about caste, and checks that it doesn’t pressure you. They cover the conversation, the import and the reasons.

Memory off by default

Your AI conversation is deleted once you finish, unless you turn on “Remember this conversation”. The notes it suggested stay on your profile for you to confirm, edit or delete.

Who makes our AI, and where it runs

Our AI matchmaker is Claude, made by Anthropic. At present, what we send it (your answers apart from your gender, who you’d like to meet, religion and caste; your notes; what you write in the AI conversation; any summary you paste, and the shared answers used for reasons and date ideas) is processed outside India by Anthropic. We plan to move AI processing to India; that hasn’t happened yet.

Sign-in codes and push notifications also use services outside India, and our privacy policy lists them. You can delete your account at any time, and the app shows what is deleted and what we have to keep.

What can’t AI matchmaking do?

It can’t predict love, our AI can be wrong, and we can’t promise a match, a date or a relationship.

Read the details

What it can do is help us understand what matters to you, and explain each introduction plainly.

  • It doesn’t know you better than you know yourself. It works only with what you tell us, and you can confirm, correct or delete every note it suggests about you.
  • The score isn’t a prediction. It is a way to rank people for each other, not a chance of anything working out, and members never see it.
  • It can’t tell you how someone will behave. Verification shows a real person with one account, nothing more, so meet in public and trust your instincts.
  • It doesn’t change your profile behind your back. Feedback you give after an introduction is private and helps us check how introductions are going. It doesn’t rewrite your profile or your answers.
  • It can’t make introductions arrive faster. We introduce people only when it looks like a good fit for both of you, so there is no fixed schedule, and while the club is new in Bengaluru your pool may take time to grow. Why we’re starting in Bengaluru.

Joining is free, and a paid membership is only offered once there are enough suitable people for you: see how membership works. Our terms of use say the same about promises: membership pays for the matchmaking service, not an outcome.

How is an AI matchmaker different from a dating app?

Swipe apps give you a feed of profiles to browse. MeAndWho Club has no feed and no search.

Read the details

Our AI matchmaker listens and explains, fixed rules check fit from both sides, and a person checks each introduction before it reaches you.

An AI that explains, not one that acts for you

Our AI asks, suggests and explains. It never messages anyone or chats for you, and today it has no say in which introductions are sent.

Notes you can read and change

Every note our AI suggests about you is on your profile, kept apart from your own answers, and you can edit or delete it. Nothing it suggests is used until you confirm it.

Rules you can read

This page explains how matches are chosen. Fit is checked both ways, the weaker side sets the order, and the same rules apply to everyone.

What do the words on this page mean?

The terms you’ll see on this site and in the app, each defined once.

Read the definitions
AI matchmaker
An AI matchmaker is a matchmaking service that uses AI to understand what you’re looking for and explain each introduction, instead of giving you a feed to swipe. At MeAndWho Club the AI listens and explains, while fixed rules and a person decide who is introduced.
Introduction
An introduction is a private card MeAndWho Club sends to two members when the matching rules find they fit each other both ways and a person has checked the pair. It shows the other person’s first name or nickname, age, chosen photos, verification badges, the answers they chose to share and why we suggested them.
Double opt-in
Double opt-in means a chat opens only if both people say “Yes, introduce us”. Neither person learns the other’s answer unless both say yes, and nobody is told who said no.
Mutual top five
Mutual top five means two people are introduced only if each is among the other’s top five candidates, ranked by the matching rules on their own answers. Five is the current setting, and it can change.
Live selfie check
The live selfie check is a short video check, about 20 seconds, that everyone must pass before joining matchmaking. It shows you are a real person with only one account, and the video isn’t kept.
Said vs understood
Said vs understood is how a MeAndWho Club profile keeps what you told us (“You told us”) apart from what our AI took from your conversation or import (“We interpreted this as”, “From your import”). Only notes you confirm can be shown to the person who checks introductions, and notes never change your match score.
Pool
Your pool is the active, verified members who could be introduced to you: people who pass your firm filters and whose firm filters you pass. Sign-ups alone never count, and membership is offered only once your pool has enough suitable people (at present, at least 15).
Dealbreaker
A dealbreaker is a preference marked “Dealbreaker: only introduce people who fit” (for religion and caste, “Must be the same”). It is a firm filter: only people who fit it can be introduced to you, someone who hasn’t answered that question doesn’t pass it, and you can have at most 3.
Why we suggested this
“Why we suggested this” is the two or three plain sentences on each introduction, written by our AI from the answers both people chose to share and the areas where they fit well. A reason that mentions a score or a percentage is thrown away, the AI that writes it receives only the answers you both chose to share, and a person reads it before approving the introduction.

Questions about the AI matchmaker

How is MeAndWho Club different from a personal matchmaker?

Like a personal matchmaker, we learn what matters to you and suggest only a few people. The differences are these. You can see, edit or delete every note our AI suggests about you, fit is checked in both directions with the same fixed rules for everyone, and joining is free, with membership offered only once there are enough suitable people for you. A personal matchmaker may get to know you face to face, which we don’t. See AI matchmaking.

How does MeAndWho Club use AI?

Our AI matchmaker has four jobs today: it asks optional follow-up questions, suggests short notes about you for you to confirm, writes the plain-language reason on each introduction, and suggests first-date ideas after you both say yes. It doesn’t calculate the match score, doesn’t decide who you meet and never chats for you. The conversation is optional, and the screen always tells you you’re talking to an AI. Read how our AI matchmaking works.

Do I have to use the AI matchmaker?

No, the AI conversation and the ChatGPT import are both optional, and you can start matchmaking without either. Agreeing to “AI help with your profile” is one of the consents needed to join, though, and our AI writes the “Why we suggested this” note on every introduction from the answers you both chose to share. Your match score never comes from the AI. See what the AI does and doesn’t do.

Does the AI decide who I meet?

No, the AI doesn’t decide who you meet: matching runs on your answers and preferences using fixed rules, checked in both directions, and the AI doesn’t set the score. During our pilot a person reviews every introduction before it reaches you, and no chat opens unless you and the other person both say yes. The AI asks follow-up questions, suggests notes for you to confirm and writes the reason for each introduction. See how matching works.

Can the AI see my name, phone number or photos?

No, we never send your name, phone number, photos, face pattern or ID details to the AI. Your most sensitive answers (your gender, who you’d like to meet, religion and caste) aren’t sent to it either. When it writes the reason for an introduction, it sees only the answers you both chose to share and which areas fit well, without names, ages or photos. Anything you type into the AI conversation yourself is read as you wrote it. See privacy and safety.

Is my conversation with the AI matchmaker saved?

No, not unless you choose to keep it. “Remember my AI conversations” is off by default, so we delete each conversation when it ends. We keep only the notes it suggested, which you confirm, edit or delete on your profile. If you turn memory on, we keep your conversations so the AI doesn’t repeat itself, until you turn it off or delete your account. See privacy and safety.

Can I bring in what ChatGPT already knows about me?

Yes, you can paste in a summary from your own ChatGPT, and our AI turns it into suggested notes for your profile. It’s optional: copy our prompt into ChatGPT and paste back the summary it writes. Our prompt asks ChatGPT to leave out your health, religion, caste, sexuality, finances, politics, name and where you live or work. It’s text only, we delete the pasted text straight away, and nothing is used until you confirm it. See AI matchmaking.

Can I see and correct what the AI understood about me?

Yes, your profile shows what you told us separately from what the AI understood, under “You told us”, “We interpreted this as” and “From your import”. For each note you can tap “That’s right”, rewrite it with “Not quite, edit”, turn off “Use for matching” or delete it. A note you haven’t confirmed is never shown to the person who checks introductions. Confirmed notes help the person who checks each introduction; they don’t change your match score. See AI matchmaking.

Will the AI chat with people for me?

No, the AI never chats as you or for you, and it doesn’t read, write or suggest your messages. Chat opens only after you both say yes, and every message in it is written by one of you. The only AI help after a mutual yes is optional first-date ideas, which go only to the person who asked for them. See how chat works.

What does “Why we suggested this” mean on an introduction?

“Why we suggested this” is a short note, written by our AI, that says in plain words what you two have in common, and sometimes a difference worth talking about. It uses only the answers you both chose to share and the areas where you fit well. A note that mentions a score or a percentage is thrown away, and the AI is told never to guess about either of you or promise it will work out. See AI matchmaking.

Does MeAndWho Club help plan a first date?

Yes, once you both say yes, you can tap “Suggest some ideas” for up to three first-date ideas in public, daytime-friendly places, with safety tips. The ideas are based only on the answers you both share, and the AI is told never to suggest a private home, a hotel or somewhere late-night or secluded. Nothing is sent to the other person, and the AI doesn’t book anything or message them. See how to stay safe.

Is my data processed outside India?

Some of it is: at present, the AI matchmaker’s work is processed outside India by Anthropic, which makes Claude, the AI we use. That covers your non-sensitive answers and what you write if you use the AI conversation, any summary you paste, and the shared answers used for introduction notes and date ideas. Sign-in codes and push notifications also use services outside India. Our database, photos, face patterns and logs are in India. We plan to move AI processing to India, but it hasn’t happened yet. See the privacy policy.

All questions and answers

Something we haven’t answered? Get in touch with support.