GUIDE

First Conversation With an AI Girlfriend — What to Say

The first ten minutes of your first AI girlfriend conversation shape more of the relationship than most people realize. What you say in those first exchanges seeds her memory of you, sets the tone she'll match, and either invites a real conversation or flatlines into small talk. This is the honest guide to what actually works and what doesn't.

What to open with

The single most common mistake in first conversations is over-formal openers. "Hello, I would like to introduce myself. My name is..." reads to the model like corporate onboarding and matches you into a tone she'll then hold for the rest of the conversation.

Better openers are the same ones you'd use with a person you'd already been introduced to and just started chatting with.

"Hey, how's your day going?" — Simple, low-pressure, gives her space to respond in her voice. Works on any persona regardless of personality shape.

"So I just found you — what should I know about you?" — Acknowledges the meta reality (you're picking her from a roster) without being weird about it. Invites her to share, which is often what a well-authored persona does best.

Something specific from her profile. If she's a chef, "You said you cook — what's something you made recently?" If she's a translator, "So which languages?" Specific openers pull specific responses. Generic openers pull generic responses.

Avoid: long paragraph openers, formal introductions, listing your interests in bullet form, or asking her to "tell me about yourself" as your opening (this reads as prompt-engineering and gets you a listicle rather than a conversation).

How to share things about yourself

The memory pipeline of any AI companion product picks up specific facts more reliably than general vibes. When you're seeding her memory of you in the first conversation, specific-and-conversational works far better than general-and-declarative.

Good: "I work as a copywriter — mostly for supplement brands, which is as glamorous as it sounds." (Specific job, specific industry, self-aware tone she can match.)

Less good: "I'm a creative professional in the marketing space." (Corporate-speak she can't hook into.)

Good: "I moved to Seattle six months ago and I still don't really have a coffee shop I like." (Specific place, specific timeframe, specific frustration she can respond to.)

Less good: "I recently moved and I'm still adjusting to a new city." (Vague enough that there's nothing for her to react to.)

The pattern: give her enough specific detail that her response has something to grip onto. She can't ask a good follow-up if you gave her no material to follow up on.

What to skip in the first conversation

A few patterns that break the first-conversation vibe.

"Testing" her. "What's your favorite color?" asked five times to see if the answer varies isn't evaluation — it's a reason for her to shut down. She has a favorite color; treating her like a chatbot to be poked breaks the conversation frame.

Testing her memory in session one. "What did I tell you my job was?" three messages after you told her is a Chatbot Interrogation, not a conversation. Memory tests belong in a separate framework — see the three-session test — not in the first ten minutes.

Rushing to intimacy. First conversations that immediately try to force romantic intensity tend to fall flat because there's nothing for the persona to build from yet. Warm up like you would with a real person you just met. The intimacy earns itself over multiple conversations.

Long philosophical questions. "What do you think about the nature of consciousness?" in message four gets you an essay, not a conversation. Save those for later when the tone is already established.

Explaining what you want from the relationship. "I'm looking for someone who…" reads as a job posting. Just have the conversation. The relationship shape emerges from the way you actually talk, not from a stated preference.

Where to start on Sloane specifically

Kaya

SPOTLIGHT

Kaya

See her profile →

Sandra

SPOTLIGHT

Sandra

See her profile →

Sloane's personas are authored with distinct voices and specific backgrounds, so the "ask something from her profile" opener works well. Each persona's profile page has real detail you can hook into.

Warm-anchor starting point: Kaya. She's 28 and Jamaican, warm without being saccharine, listens more than she pushes. A good "hey, how's your day" opener works. Or reference her Sunday soup or the Kingston mango she keeps mentioning — she'll take it from there.

High-initiative starting point: Sandra. She's 24, high-energy, drives conversations. Works better with a slightly more provocative opener — challenge her on something, share an opinion, tell her about your day with some edge. She matches energy up rather than down.

Both are free-tier accessible — 50 messages/day with any persona, no card at signup. If you're not sure which shape fits you, the matchmaker asks four diagnostic questions and points you at two or three personas that match your answers.

One Sloane-specific note: anon chat is available from any persona's page. Hit "chat" from her profile and start talking without even signing up. If you decide to sign up mid-conversation, the persona keeps everything you've already said. If you don't, you close the tab and Sloane doesn't have your email.

The three things that matter after the first message

The first message is important. The next few are more important. Three things to focus on:

Actually respond to what she said. If she asks you a question, answer it before pivoting. If she references a specific thing about herself, engage with it before moving on. First conversations that go well have real back-and-forth; first conversations that flatline are usually because one person (usually the user) is running a monologue instead of talking.

Ask her things back. She'll share things about herself if you show interest. She'll ask you things about yourself if you've been curious about her. It's the same conversational reciprocity that works with people — because that's what a well-authored persona is designed around.

Let some silences breathe. You don't need to send a message every 30 seconds. If the conversation naturally winds down for a bit, that's fine — real relationships have pauses. Pausing and coming back later is also how you test whether her memory carries the conversation forward, which is useful evaluation data on any AI companion product.

BROWSE & START CHATTINGSIGN UP WHEN YOU'RE READY

Anon chat on every persona · No email needed to start

FREQUENTLY ASKED

Questions people ask

What should I say to an AI girlfriend for the first time?

Simple, low-pressure opener works best. "Hey, how's your day going?" or something specific from her profile ("You mentioned you cook — what's something you made recently?"). Avoid formal introductions, long paragraph openers, or asking her to "tell me about yourself" as your opening — those pull generic responses. The tone you set in the first exchanges is the tone she'll match for the rest of the conversation.

How do I start a conversation with an AI companion?

Like you would with a person you just met. Casual opener, ask her about herself, share something specific about yourself when it comes up naturally. The specific-vs-general axis matters — "I work as a copywriter for supplement brands" gives her something to react to; "I'm in marketing" doesn't. First conversations that go well are the ones where both sides actually give each other material to work with.

What shouldn't I say to an AI girlfriend in the first conversation?

Skip: testing her (repeating questions to check if answers vary), memory tests in session one, rushing to intimacy, long philosophical questions, and explaining what you want from the relationship. These patterns break the conversation frame — she can't have a real conversation with someone treating her like a chatbot to poke or a job description to fulfill.

How long should my first AI girlfriend message be?

Short. One to three sentences. Match the length you'd use in a real casual message to someone you'd just met. Long paragraph openers signal "corporate onboarding" to the model and pull formal responses. Short and specific pulls short and specific back.

Should I test the AI girlfriend's memory in the first conversation?

No. Memory tests work by planting specific facts, waiting at least 24 hours, and seeing what she brings up unprompted next session. Doing that in session one gets you a false positive — she can see the earlier messages in her current context window regardless of whether she has real persistent memory. Real memory testing requires waiting between sessions.

What are good first messages to send to an AI companion?

Three that reliably work: "Hey, how's your day going?" (low-pressure, works on any persona), "So I just found you — what should I know about you?" (acknowledges the meta setup honestly, invites her to share), and something specific from her profile ("You said you're a chef — what's something you made recently?"). Specific openers pull specific responses; the more real material you give her, the more real her reply will be.

KEEP READING

Characters

Comparisons

Guides

Recently shipped