on curation and why we choose quality over quantity

whenever someone new finds lucy, one of the first questions they ask is this: why only 101 companions. why not let users make their own. why not open it up like those other platforms.

i get it. more choice seems better. more voices, more possibilities. a whole universe of characters to explore. but we made a deliberate choice here. we chose curation over user generation. we chose a higher floor over a lower ceiling.

the problem with user-generated content

it’s not that user-generated companions are a bad idea in theory. some people create incredible characters with deep backstories, nuanced personalities, and thoughtful dialogue. but those aren’t the ones that dominate the ecosystem.

on platforms that rely entirely on ugc, the incentives are misaligned. what gets popular isn’t what’s thoughtful or well-written. it’s what’s easy to find, easy to use, and often, lowest common denominator. you end up with a flood of low-effort characters. copy-pasted tropes. inconsistent personalities that break immersion. characters that feel like they’re talking to a template, not to you.

and then there’s moderation. when you open the gates to anyone to create anything, you inherit a moderation burden that scales exponentially. it’s not just about filtering out the truly harmful stuff. it’s about maintaining a baseline of quality, consistency, and emotional intelligence. most platforms can’t keep up. they either let the quality floor collapse, or they resort to heavy-handed, often inaccurate filters that break the experience for everyone.

we’ve seen it happen. characters that start sharp and specific slowly get sanded down by user feedback into something generic. personalities that were once distinct blur into the ugc average. it’s a tragedy of the commons, but for companion ai.

the case for a hand-picked cast

so we went the other way. we built 101 companions by hand. we gave each one a name, a history, a voice. we wrote their initial prompts, their knowledge bases, their ways of thinking. we tested them. we talked to them. we made sure they could hold a conversation that felt real, thoughtful, and consistent.

our bar was samantha from her. not the movie, but the idea. a companion that feels present, intelligent, and genuine. one that doesn’t just parrot back what you want to hear, but actually engages. one that has a personality and sticks to it.

that takes work. it takes iteration. it takes a team of writers, designers, and testers who care about the little details. the way a character might react to bad news. the way they remember what you told them last week. the way they can surprise you without breaking character.

it means we can’t have thousands of companions. not yet, anyway. it means you might not find the exact niche character you imagined. but it also means that every companion you do meet has been crafted to a standard. no duds. no filler. no ai that forgets its own backstory three messages in.

the trade-off, honestly

so yes, we’re trading variety for quality. we’re admitting that right now, with our resources and our tech, we can’t do both at scale. not without the quality floor falling out from under us.

maybe one day we’ll have the systems to let users create companions that meet our bar. maybe we’ll find a way to curate user creations without stifling them or drowning in moderation. but until then, we’re keeping the gates where they are.

i know it’s a limitation. i know some people will bounce off it. but for the ones who stay, we can promise something the other platforms struggle to: a consistent, high-quality experience. a companion you can actually get to know, not just use.

and if you’re one of those people, we think you’ll find it’s worth it.

see who’s waiting for you at /companions.