The Series A, led by Seligman Ventures, takes total funding past $21M and funds an "asynchronous" voice architecture built to listen, reason, and speak in parallel — the way people actually talk.
2026-07-31 · by Moe Ameen
Smallest.ai, a voice-AI company founded in late 2024 by CEO Sudarshan Kamath, announced on July 30, 2026 that it raised a $13M Series A led by Seligman Ventures, with Sierra Ventures and 3one4 Capital participating. The round brings the company's total funding to over $21M. The pitch is narrow and deliberate: not a general-purpose model, but voice specifically — accents, dozens of languages, and staying intelligible in noisy, real-world audio.
Alongside the raise, the company introduced Voice 4.0, an architecture it describes as "asynchronous." Where most voice systems wait for a full conversational turn before responding, Smallest.ai says Voice 4.0 listens, reasons, and speaks in parallel — the way a person is already forming a reply, and might interrupt, while you're still talking. Its Hydra speech-to-speech model is the first product built on it, aimed at near-zero response lag in live conversation.
The company's existing lineup sits underneath that: Lightning and Waves for text-to-speech (Waves offers voice cloning from a few seconds of audio across 30+ languages), Pulse for speech-to-text (the company says 38 languages with millisecond-level latency, plus diarization, emotion detection, and redaction), and Atoms, a real-time voice-agent platform that plugs into business phone systems. Named customers include RingCentral and Truecaller. Coverage placed Smallest.ai against voice-AI incumbents including ElevenLabs and Cartesia, and regional players like Sarvam. The core market is enterprise voice agents and phone experiences, not content creation — but the same TTS models double as fast, cheap narration for anyone making video.
The takeaway for creators isn't the funding number — it's that realistic voiceover is now a cheap, fast input, and the leverage is entirely in what you build on top of it. Kompozy is that layer, and the two split cleanly at the script. Kompozy generates the on-brand script — held to your Persona Brief so it's already audience-fit — that you run through Smallest.ai's Waves (a cloned voice for a consistent host, or a stock voice in another language) to produce the standalone narration.
From that same script, Kompozy produces what the raw audio can't become on its own: Clipped and Persona Shorts for the feeds (its talking-head video renders through HeyGen's own native voice), a Carousel, Quote Graphics, a Blog Article, an Email Newsletter, and Text Posts — all governed by your Persona Brief so the written voice matches the spoken one — and, via Autopilot and a per-post review pipeline, reframes and publishes across nine destinations: the eight primary social platforms plus blog and email. The news is that voice got commoditized; the response is to turn one script into a week of on-brand content and ship it everywhere — which is what Kompozy does.
Smallest.ai announced a $13M Series A on July 30, 2026, led by Seligman Ventures with Sierra Ventures and 3one4 Capital participating. The round brings its total funding to over $21M.
Voice 4.0 is the architecture Smallest.ai introduced with the raise. It calls it "asynchronous": the system listens, reasons, and speaks in parallel rather than waiting for a full conversational turn, to cut response lag in live voice. Its Hydra speech-to-speech model is the first product on it.
Yes. The company targets enterprise voice agents, but its Lightning and Waves text-to-speech models produce fast, realistic narration and support voice cloning across 30+ languages — directly usable as a standalone voiceover for a video. Kompozy complements it by writing the on-brand script you voice and generating and publishing the video and posts around it across nine destinations.