How we think
The best interface isn't a screen — it's a conversation.
We're convinced the next leap in digital products is voice: AI that hears what people really mean, and answers with a presence of its own. Here's how we think about speech, sound, and the products they make possible.
What we believe
Voice is the oldest interface — and, suddenly, the newest.
Long before screens, we talked. It's still the fastest, most human way we trade ideas: full of nuance, timing, and intent. The tools to give software that same fluency have finally arrived — and we think they quietly change what a product can be.
Speech-to-text
Listening is more than transcribing
Good speech-to-text doesn't just turn sound into words — it catches how something was said: the emphasis, the pause, the half-formed thought. We design for understanding, not dictation, so a system hears meaning and not just phonemes.
Text-to-speech
A voice is a presence, not a playback
Synthetic voices used to announce that they were machines. They don't have to now. We treat text-to-speech as identity — pace, warmth, character — so a product doesn't talk at people, it talks with them.
Put the two in a loop, and you get conversation.
Fast listening, expressive speaking, and a capable model in between — that's not a feature, it's a new kind of interface. No menus, no syntax; just talk. We think the products that define the next decade will be the ones you can simply speak to.
We build what we believe
There's a sona on this page. Talk to it.
The sona in the corner is an interactive AI you can genuinely hold a conversation with — a working example of everything above. Open the bubble, say hello, then tell us about a product that should have a voice.