Smallest.ai has raised $13 million in Series A funding to develop voice AI that responds quickly enough to sound more like a human conversation. The startup argues that conventional large language models can introduce noticeable pauses in spoken exchanges, so it is building smaller, specialised models that listen, process and respond simultaneously; this could make AI customer support agents feel more natural.

The round was led by Seligman Ventures, with Sierra Ventures and 3one4 Capital participating, bringing Smallest.ai’s total funding to more than $21 million. Its model handles real-time, voice-specific challenges including accents, multiple languages and noisy environments, while referring unfamiliar questions to a larger model; customers include RingCentral and Truecaller, and competitors include ElevenLabs and Cartesia.