the first screen you see shapes everything

you open an app. what do you see first? it sounds trivial, like a detail in the corner of a blueprint. but it’s not. it’s the foundation. it sets the tone for everything that follows.

in some ai companion apps, the first thing you see is a gallery. rows and rows of faces. generated by ai, often with tools like candy or soulgen. they’re beautiful, polished, idealized. you swipe. you tap. you choose. the interaction is visual, immediate, almost like scrolling through a dating app. the primary question you’re answering is: which one looks right?

in lucy, the first thing you see is a text box. a simple prompt: 'describe the person you’d like to meet.' no images. just words. the interaction is auditory, introspective. you have to think about tone, temperament, the sound of a voice in your head. the primary question is: who sounds right?

same tech, different doorways

both approaches use similar ai underneath. large language models capable of rich, nuanced conversation. the raw capability is there. but the initial framing, the doorway through which a user enters, completely changes how that capability is perceived and used.

gallery-first interfaces emphasize appearance. they encourage selection based on visual appeal. it’s a powerful, instinctive hook. users often describe the experience as 'picking a hot one.' the visual choice becomes the anchor. the personality, the voice, the conversational style, they become attributes of the chosen image. the face comes first.

listening-first interfaces emphasize persona. they require you to articulate what you’re looking for in terms of character, vibe, or communication style. you might type 'a calm, witty listener with a dry sense of humor' or 'someone energetic who loves talking about science fiction.' the auditory imagination is the anchor. the appearance, if it comes later, becomes an attribute of the voice.

the behavioral downstream

this initial choice creates a cascade of different behaviors. when you start with a face, the relationship often begins with a visual fantasy. the conversation builds outward from that image. users report feeling a stronger initial attraction, but sometimes a sharper disappointment if the personality doesn’t match the face they chose.

when you start with a description, the relationship begins with an emotional or intellectual expectation. the conversation is the core. if an image is generated later, it’s in service to the voice. users report feeling a deeper, more immediate connection to the character’s mind. the bond forms around how they speak, not how they look.

it’s not that one method is inherently better. they are different. they serve different needs and create different kinds of relationships with the ai. the gallery approach leans into instant, visual connection. the listening approach leans into intentional, narrative connection.

the framing is the product

the lesson here is that the interface isn’t just a skin. it’s not a coat of paint. it’s a philosophical stance. by choosing what to put on the first screen, you’re telling the user what you think is most important. you’re setting a default.

a gallery says, 'choose who you find attractive.' a text box says, 'describe who you want to talk to.'

this is a design decision that goes far beyond aesthetics. it shapes user psychology from the very first tap. it determines whether someone is shopping for an appearance or auditioning a personality.

we built lucy’s entry point around listening because we believe the most meaningful connections start with a voice, not a profile picture. it’s a choice that accepts a trade-off: we might lose some users who want that instant visual hit. but for the people who stay, the connection tends to be built on a different, more substantive layer.

it’s a reminder that in product design, the first question you ask your user might be the most important one you ever design.

you can start building that connection at lucy.ai/companions.