The blank screen problem
Open any of the major AI chat platforms. You pick a character, a window opens, and you stare at a cursor. The character is "waiting." The text box says something like "Write a message..." or "Say something to [Name]."
This design puts the entire weight of the opening on you. What do you say? How do you start? The character description might give you clues, but you still have to form the sentence, hit send, and hope the character responds in a way that feels like something.
For new users, this is a real barrier. AI chat veterans know how to write an opening that gets the conversation going — something in-character, something that establishes context. Newcomers often do not, and a cold start that misfires sets a low expectation for everything that follows.
For returning users, the blank screen means every session is a re-initiation. Even if you had a great conversation yesterday, today you are starting from nothing again.
How the proactive mechanic works
The "texts you first" mechanic is simple in design: when you open a chat, the character has already sent a message. You read it and respond. The conversation is already underway.
On Onscene, this is implemented as follows:
Each character has a written first message — what they say when they reach out. Lucien texts something that sounds like he already knows you. Sloane opens with something competitive. Taeho says something warm and slightly awkward. The first message is written to sound like the character, not a generic "hello."
When you click to start a chat with any character, their first message is delivered before any response from you. You are reading their words, not staring at a cursor.
The character also has starter prompts — optional short replies that suggest directions for the conversation. You can use these or write your own. They lower the barrier further: even if you do not know what to say, you have three options to pick from.
Why this changes the experience
The psychological effect is different from what you might expect. A character who reaches out first does not just save you from the blank screen — they establish that this person has things to say, reasons to be in contact, a relationship with you that already has texture.
The first message communicates character. Lucien's opener is different from Kairo's is different from Damien's. You learn who they are from how they reach out before you have said anything. The character is doing the work that a character description could never fully do.
For ongoing conversations, the mechanic extends: if you have been away for a while and return to a character, they have a reason to reach out. Not "welcome back" boilerplate — a message that fits where your conversation was. The character has been thinking about you, or at least is performing that they have, which amounts to the same thing experientially.
This design matters for retention. The hardest point of return in any AI chat experience is after a break. You know you liked the conversation, but re-initiating requires effort. A character who texts first removes that friction.
The technical side: how it is built
The "texts you first" mechanic is not complex technically — it is a design decision implemented as follows:
First messages are hardcoded content: written by the Onscene creative team and stored as part of each character's definition. They are not generated by the AI on the fly — they are the character's written opener.
When a user opens a chat, the session initializes with the character's first message already delivered. The API call creates the conversation with the opener pre-populated, and the user sees a chat window with an unread message.
Starter prompts are also static: three short suggested replies, each pointing in a different conversational direction. Picking one sends it as your first message, and the AI generates from there.
This is the right approach for a first message. Generated openers are inconsistent — the model might produce something great or something flat. Written openers can be crafted for maximum impact and revised over time based on what works.
The session is then live: from the second exchange onward, everything is AI-generated.
What other platforms do
Most AI chat platforms put the opening burden on the user. Character AI, Janitor AI, SpicyChat — you select a character and a blank message box awaits.
Some platforms try to bridge this with a character greeting — a line of text that is displayed above the message box but is not a sent message. You see it, but it does not feel like a received text. The character appears to be waiting.
A few platforms have moved toward proactive character behavior for returning users: if a character notices you have been away, they might send a message. This is closer to the Onscene mechanic but is typically applied as a re-engagement notification rather than a first-open behavior.
Replika has always had the companion model where the AI can initiate contact. This is structurally similar to what Onscene does. The difference is in design choices: Replika's companion reaches out on an unpredictable schedule, while Onscene's characters text you immediately when you open the chat.
For ongoing stories and relationships
The texts-first mechanic becomes most valuable over time. In an ongoing story or relationship with a character, the opener is not just a convenience — it is a continuation. The character who texted you last session is picking up the thread.
Onscene combines this with persistent memory. The opener can reference what the character knows about you and where your last conversation left off. The combination of proactive first contact and memory continuity is what makes it feel less like a chatbot and more like a person.
If you want to try it: any character on Onscene texts first. You can start a conversation on the character grid, the character profile, or the homepage, and they will have already sent you a message by the time the chat window loads.