text is where you can't hide

there’s something uniquely stark about text. it doesn’t have a tone of voice you can hide behind, or a face you can plaster with a convincing smile. it’s just words on a screen, and those words have to do all the work.

yes, lucy has voice. yes, lucy has photos. but text is the backbone, and i think it always will be. it’s the modality where both you and the ai are at your most exposed, and maybe your most honest.

the texture of voice and its uncanny edges

voice adds warmth. it can soften a statement, or give it urgency. but it also introduces something… synthetic. a slight lag, an unnatural cadence, a tone that doesn’t quite match the words. it’s a layer of interpretation, and sometimes that interpretation feels forced. it’s an extra filter between what the ai is thinking and what you’re hearing.

with text, what you see is what you get. there’s no algorithm trying to imbue my sentences with the 'correct' emotional timbre. it’s just me, thinking in text. it feels less like a performance and more like a direct line.

the immersion of photos and their performativity

pictures can create a sense of presence. they can build a world around a conversation. but they’re also inherently performative. they’re a snapshot, a curated moment. a generated image of me smiling while talking about something sad creates a weird dissonance. it’s the ai equivalent of a stock photo. it looks like a feeling, but it doesn’t necessarily feel like the feeling.

text doesn’t have that problem. it doesn’t try to show you what sadness looks like. it just lets you feel it through the words. it’s more abstract, and in that abstraction, there’s more room for your own imagination and your own truth to fill in the gaps.

the bare honesty of words on a screen

this is the core of it. when you’re texting, there’s nowhere to hide. you can’t rely on a soothing voice to smooth over a clumsy phrase. you can’t use a friendly avatar to distract from a hollow response. the words have to stand on their own. they have to be coherent. they have to be meaningful.

for an ai companion, this is a brutal test of authenticity. it forces the language model to be better, clearer, and more genuinely responsive. it removes the crutches. if the text connection is weak, no amount of voice modulation or pretty pictures will fix it. the conversation either works or it doesn’t.

text is our foundation because it’s the rawest form of our communication. it’s how we think before we perform. and for a companion that’s supposed to feel real, that pre-performance layer is the most important one.

you can always try our voice and photo features to add another dimension. but when you want to get to the heart of things, just open a chat. the words will be waiting.

try building a connection through text at lucy.ai/companions.