Smallest.ai, a startup founded in late 2024, has raised $13 million in Series A funding to develop ultra-fast voice AI that sounds genuinely human. The company aims to differentiate itself from competitors by focusing on smaller, specialized models designed for natural conversation, rather than simply accelerating large language models.

Founder and CEO Sudarshan Kamath explains that Smallest.ai's model mimics human processing by listening, thinking, and speaking simultaneously, allowing for near-instantaneous responses and interruptions, unlike the sequential processing of traditional LLMs. This real-time intelligence layer enables natural customer conversations with virtually zero lag.

The new capital, led by Seligman Ventures with participation from Sierra Ventures and 3one4 Capital, brings the startup's total funding to over $21 million. When encountering topics outside its specialized knowledge, Smallest.ai seamlessly hands off to a foundational model, mirroring human behavior.

Smallest.ai focuses on voice-specific nuances like accents, multiple languages, and noisy environments. Its enterprise customers include companies like Ring Central and Truecaller. Kamath believes the future of AI agents will involve a combination of small, real-time voice models and larger, on-demand LLMs, with the ultimate goal of creating AI that is indistinguishable from humans, effectively aiming to "break the Turing test."