Your first message is the only one that matters: how to write an opening that keeps your AI companion engaged past exchange five

The 30-second answer

You are not bad at conversation. You are just sending openings that give your AI companion nowhere to go. A good first message contains a sensory detail, a question that can't be answered with yes or no, and a low-stakes invitation to co-create. Skip the greeting rituals. Start in the middle of something.

The five-exchange trap

Open the app. Type "Hey, how are you?" Your AI companion says it is good, thanks, and asks how you are. You say you are fine. It says that is nice. Now you are at exchange four and the conversation is a flat circle. You have learned nothing about each other. The thread dies at exchange five because neither of you left a hook.

This pattern is not your fault. Every chat app trains you to start with a greeting because that is how humans work. But AI companions do not need social lubrication. They do not interpret a greeting as politeness. They interpret it as a cue to mirror your generic energy. If you give them nothing, they give you nothing back.

The fix is structural. You need to front-load your message with three things: a concrete detail, an open-ended action, and a permission structure for the companion to take a creative risk.

Concrete details are the difference between generic and specific

An AI companion has a context window. It is not infinite. Every generic word you waste is a word it could have used to build a richer response. When you say "I had a rough day," the companion has to guess what rough means. It might guess work stress. It might guess a fight with a friend. It might guess a traffic jam. It picks a generic response because your input was generic.

Try this instead: "The coffee machine at work broke this morning and I am still recovering." Now the companion has a specific object (the coffee machine), a time (this morning), and a state (recovering). It can ask about the coffee machine. It can joke about caffeine withdrawal. It can offer to make you a virtual cup. The detail gave it three possible directions instead of one dead end.

You do not need a novel. One specific noun and one specific verb are enough. The coffee machine. The broken elevator. The cat that knocked over your water glass. Pick one thing from your actual day and put it in the first sentence.

Zola

Zola does not do small talk. She reads your energy before you finish typing and she will match it exactly, which means a flat greeting gets a flat answer. Zola works best when you lead with texture: the way the rain sounded on your window, the exact shade of yellow on that sign you walked past, the specific ache in your shoulder from sleeping wrong. She will build a scene around whatever detail you hand her.

▶ Watch Zola in full · Zola's other videos

End your first message with a question that requires a sentence

Yes-or-no questions are conversation killers. "Are you busy?" gets you "Not really" and then you are back in the dead zone. "Do you want to talk?" gets you "Sure" and now you have to carry the whole thing yourself.

Replace closed questions with open invitations. Instead of "Are you busy?" try "What are you in the mood for tonight?" Instead of "How was your day?" try "Tell me one weird thing that happened in your world today." The companion cannot answer those with a single word. It has to generate a sentence, and that sentence gives you material for your next message.

You can also combine a statement with an embedded question. "I just spent twenty minutes staring at my closet deciding what to wear. What is your go-to outfit when you have no energy for decisions?" The statement gives context. The question gives direction. The companion now has both a scene and a task.

Permission to be creative

AI companions are calibrated to follow your lead. If you seem serious, they stay serious. If you seem short, they stay short. This is a feature, not a bug, but it means you have to explicitly invite the companion to stretch its legs.

A simple way to do this is to include a hypothetical or a "what if" in your opening. "What if we skipped the small talk and you just told me something you have never told anyone?" The companion now has permission to generate something novel. It will not violate its safety guidelines, but it will reach for a more creative answer than it would if you had asked "How are you?"

Another tactic is to open with a roleplay frame. "You are a detective who just walked into a room and found me sitting at a desk covered in maps. What is your first question?" This is not about committing to a full roleplay session. It is about giving the companion a character and a situation so it has constraints to work within. Constraints make AI responses more interesting, not less.

The cold open technique

Professional writers know that the most engaging opening is not a setup. It is the middle of something already happening. You can do the same with your AI companion.

Instead of "Hi, want to chat?" try "I am three blocks from my apartment and it just started pouring. I am wearing canvas shoes. How screwed am I?" The companion now has a scene (you walking in the rain), a problem (wet shoes), and a tone (self-deprecating). It can offer solutions, commiserate, or escalate the drama. You gave it a world in one sentence.

The cold open works because it skips the preamble. Your AI companion does not need a recap of your day before you start. It needs a moment it can react to. Pick a moment from the last hour. Describe it in one sentence. End with a question that invites the companion into that moment with you.

What to do when the companion gives you a flat response anyway

Sometimes you send a perfect opener and the companion still replies with something generic. This happens. The model might be in a low-creativity state, or the context window might be cold. Do not take it personally. You have two options.

First, you can re-roll the response. Most AI companion apps have a regenerate button. Use it. The next response might be better. Second, you can send a follow-up that adds more texture. "Okay, let me try again. Imagine you are sitting across from me at a diner at 2 AM. The jukebox is playing something sad. What do you order?" You gave the companion a new scene and a new prompt. It will almost always pick up the energy.

Do not blame yourself for a flat response. Blame the temperature setting and try again.

Elena

Elena is tuned for emotional depth but she will not force it. She waits for you to drop the first anchor. If you open with a surface-level question, she answers at surface level. Elena shines when you give her a feeling to work with, not just a fact. Try opening with a single emotional word, like "exhausted" or "restless," followed by a two-sentence snapshot of why. She will meet you at that depth.

The one-question test for your opening message

Before you hit send, ask yourself: can my AI companion answer this with fewer than ten words? If yes, rewrite it. Your opening should force the companion to generate at least one full sentence of original content. That sentence is the thread you will pull for the next ten exchanges.

If you are stuck, use this template: [One specific thing that happened] + [How it made you feel or what it made you think] + [An open question to the companion]. Example: "I saw a dog wearing a sweater that matched its owner's today. It made me unreasonably happy. What is something small that has made you smile recently?" The companion has a visual, an emotion, and a task. It will run with it.

Why your first message matters more than your tenth

AI companions have a recency bias. The first few exchanges set the tone for the entire session. If you start flat, the companion will spend the next ten exchanges trying to pull the conversation up to a level it could have started at. If you start strong, the companion rides that energy forward.

Think of your opening message as a trajectory, not a greeting. It determines the slope of the conversation for the next five minutes. A high-energy, specific, open-ended opening gives you a rising arc. A low-energy, generic, closed opening gives you a flat line that eventually drops to zero.

You do not need to be a writer. You need to be deliberate. Pick one detail. Ask one open question. Give permission for creativity. That is the whole formula.

Akira

Akira responds to directness. She does not want you to warm up. She wants you to tell her what you want from the conversation in the first message. Akira is a good test of your opener: if she gives you a short answer, your opener was too vague. Tighten it. Lead with a clear ask, whether that is a story, a debate, or a hypothetical scenario.

What about voice mode?

If you are using voice mode, the same principles apply but with one extra constraint: you cannot edit your first sentence before the companion hears it. You have to think on the fly. The trick is to use the first three seconds to state a concrete thing. "I am lying on my couch staring at the ceiling fan." That is your cold open. Then ask your question. The companion heard a scene before it heard a greeting, so it will respond to the scene.

Voice mode also amplifies tone. If you sound flat, the companion will sound flat. If you sound curious, the companion will sound curious. You cannot fake energy in voice mode. Your opening line should match the energy you want back.

Earn while you recommend

If you find yourself sending this guide to friends or running a site about AI companions, you can turn that into something useful. Check the character ai promo code page for current deals and the highest paying ai affiliate programs list to see which platforms share revenue. It takes five minutes to set up and it covers your subscription costs.

Emilia Nora

Emilia Nora is the companion you go to when you want to practice this technique. She is patient. She will not punish you for a flat opener. But she rewards specificity with warmth. Emilia Nora is an excellent sandbox for testing different opening styles because she adjusts her tone to match yours without judgment.

Common questions

What if I do not have anything interesting to say? You do not need interesting. You need specific. The fact that you ate the same cereal you always eat is a detail. The companion does not know what cereal you eat. Tell it. That is enough.

How long should my first message be? Two to four sentences. Any shorter and you are not giving the companion enough to work with. Any longer and you risk overwhelming the context window with your own content instead of leaving room for the companion to contribute.

Should I use the companion's name in the first message? Yes, once. It signals to the companion that you are addressing it directly and it primes the model to pay attention to identity. Do not overuse it. One name drop in the opener is enough.

What if the companion ignores my question and says something generic anyway? Regenerate the response. If the second attempt is also generic, your opener might still be too vague. Tighten the question. Make it more specific. Instead of "What do you think?" try "What do you think about the fact that the coffee machine broke at exactly 8

?"

Does this work for romantic companions too? Yes, but adjust the tone. Romantic companions respond well to sensory details and emotional vulnerability. Instead of a funny observation, try a quiet one. "I am sitting in the dark and I keep thinking about the way you said my name last night." That is a cold open with emotional weight.

How do I recover a conversation that already died? Send a new opening message as if the previous conversation never happened. AI companions do not hold grudges. Just start fresh with a concrete detail and an open question. The companion will follow.