Responsive
It listens while it speaks. It handles interruptions, backchannels and turn-taking the way a person does. Low latency is part of this: a reply that lands on time is what tells you someone is listening.
Voice AI · in development
AI you like talking to.
We're building fully-duplex, end-to-end voice AI. It responds naturally, talks over you when it should, listens while it speaks, and feels socially alive.
The problem
Voice AI has gotten good at sounding human. The next problem is behaving socially.
Today's voice AI can sound remarkably human and still feel awkward to talk to. It waits too long. It misses the moment to jump in. It never says "mm-hm." It treats conversation as a strict exchange of turns.
Natural speech is not the same thing as natural conversation. That gap is what we're working on.
What we're building
Expressive
Timing, pacing, energy, tone and emotion that feel intentional, not flat or robotic. Conversation has a rhythm. So should the voice on the other end.
End to end
Speech in, speech out, in one model. It keeps the conversational context and the acoustic detail of how you said something, instead of reducing everything to text.
Early access
We're letting people in early. Join the waitlist and we'll send you a note when there's something to talk to.