Stellon Labs GTM brain

Stellon Labs builds compact AI models, including KittenTTS, for developers deploying intelligence on edge devices.

category

On-device AI models

market profile

  • text: Small models address the compute and memory constraints of phones, wearables and embedded systems.
  • text: KittenTTS gives the company an open-source speech entry point into the edge AI developer market.

personas

  • title: Embedded ML engineer
    pains:
    • Speech models exceed device memory or require a GPU
    • Cloud inference can add latency and connectivity dependence
  • title: Voice product engineering lead
    pains:
    • Need expressive speech in a small application footprint
    • Must evaluate model quality and integration effort on target hardware

icp

  • text: Teams building speech interfaces for phones, wearables, robots or embedded devices; account size is not specified.
  • text: Developers and enterprises experimenting with small models on constrained hardware.

message pillars

  • text: Position compact models for constrained hardware.
  • text: Lead with the lab's research focus.
  • text: Show the edge deployment application.

objections

  • objection: Can a tiny model meet our voice quality requirements?
    response: Evaluate KittenTTS against your own voice and device benchmarks before committing.
  • objection: Will it fit our hardware and integration constraints?
    response: Prototype on the target device and measure footprint, latency and compatibility.

competitors

  • name: Supertone
    why: Supertonic is another local speech model option.
  • name: Kokoro
    why: Open speech model used by local AI developers.

influencers

  • name: Rohan Joshi
    url: https://www.linkedin.com/in/rohan-m-joshi
    why: Co-founder sharing model releases and developer updates.

keywords

  • on-device text to speech
  • tiny TTS model
  • edge AI speech model
  • offline voice synthesis
  • embedded speech model
  • small AI models

voice

  • technical
  • concise
  • research oriented

signals

  • text: Hiring embedded ML engineers or building an on-device voice assistant suggests a likely evaluation need.
  • text: Active work on offline speech, robotics or wearables suggests hardware constraints where compact models matter.

pricing

model: unknown

motion

community_led

channels

  • GitHub
  • LinkedIn
  • X
  • Hugging Face

confidence

medium

Founders

  • name: Rohan Joshi
    title: Founder/CEO
    role: CEO
    bio: Co-founder and CEO at Stellon Labs where we train open-source frontier speech models! We're better known as KittenML (kittenml.com) Checkout KittenTTS at https://github.com/KittenML/KittenTTS Ex-Meta Research, CMU SCS
  • name: Divam Gupta
    title: Founder/CTO
    role: CTO
    bio: Packing intelligence in super-tiny AI models.

Back to YC companies