Build voice AI that speaks like people do.
We’re a small team making phone calls answerable by an AI that actually sounds like it’s from here - in more than 130 languages, in real time.
Read about Vistrow Voice and how we build, or see what we shipped recently.
Hard problems, close to real calls.
Voice is unforgiving
A chatbot can take two seconds to think. A phone call can’t. Everything here is judged on whether it feels natural at real-time latency - that constraint makes the engineering genuinely hard.
Built for how India talks
More than 130 languages, mid-sentence code-switching, and grammatical gender that has to agree turn after turn. Most global voice AI treats this as an edge case. It’s our default.
Small team, real ownership
No layers between writing something and seeing it answer a live call. If you ship it, it’s yours - including when it pages you.
No open roles right now.
We hire in bursts, and we do read speculative applications. If you’ve built something in speech, real-time audio, or Indian-language NLP, tell us what you made and how - that’s more useful to us than a CV on its own.
Send us your work