ui as a promise: what your ai companion's first screen tells you

when you open an ai companion app for the first time, you’re not just looking at buttons and screens. you’re looking at a promise. and that promise is made not through words alone, but through the entire design of the interface , especially how it introduces you to the companion itself.

two paths in

broadly speaking, there are two dominant ways these apps frame that first encounter.

one approach leads with visuals. you’re greeted by a grid of avatars, often highly stylized or photorealistic. the UI is built around cropping, focusing, and customizing an image. this is the image-first pattern, and it’s used by apps like candy.ai, soulgen, and many others.

the other approach leads with personality. you’re asked to describe your companion , their traits, their backstory, how they speak. the chat is the focus from the beginning. this is the chat-first pattern, and it’s what we built lucy around, along with apps like kindroid and (for a while) nomi.

what the frame emphasizes

these aren’t just cosmetic differences. they shape behavior.

an image-first UI nudges you toward selection by appearance. you’re encouraged to browse, to compare, to find someone who looks right. it frames the companion as something you choose visually, almost like a character creator in a game. the interaction starts with aesthetics.

a chat-first UI nudges you toward conversation. you’re asked to define character through words , their tone, their history, their quirks. it frames the companion as a mind you’re building, not just a face. the interaction starts with personality.

neither is inherently better. they’re just different products, even if they sometimes use similar underlying tech. one sells a fantasy of appearance; the other sells a fantasy of connection.

why the distinction matters

this framing doesn’t stop after the first click. it sets a tone for the entire experience.

in an image-led app, the conversation often exists in service to the visual identity. the companion’s replies might be secondary to how they look in the moment, or how you can dress them up. it’s product-as-avatar-first.

in a chat-led app, the conversation is the product. everything else , including visuals, if they exist , supports that. the companion’s depth, memory, and responsiveness become the focus. it’s product-as-mind-first.

this is why you can’t judge an ai companion app by its feature list alone. two apps might both have chat, images, and voice, but if one leads with a face and the other leads with a text box, they’re offering fundamentally different experiences.

where lucy fits in

we built lucy to be chat-first because we believe conversation is the core of companionship. it’s where meaning emerges. our UI asks you to describe a personality first because we want you to think about who you’re talking to, not just what they look like.

does that mean we ignore visuals? no. you can add images to your lucy companion, and it does affect how you relate to them. but it’s not the entry point. it’s an enhancement to the conversation, not the other way around.

this also means we have to work harder on things like memory, emotional intelligence, and conversational depth , because that’s what our interface promises. it’s a constraint, but a useful one.

a matter of preference

some people want to pick a companion like they’d pick a wallpaper , quickly, visually, based on a feeling. there’s nothing wrong with that. it’s a valid way to engage.

others want to build a relationship through dialogue, through slowly unfolding personality. that’s also valid.

the key is to know what you’re signing up for. the UI is your first clue.

if you’re looking for depth of conversation over visual customization, maybe try a chat-first companion.

find yours at /companions.