Smallest.ai Raises $13M to Build Enterprise Voice 4.0

Enterprise voice agents can answer calls at scale, but delays, awkward interruptions and disconnected models still make many conversations feel automated.

San Francisco-based Smallest.ai has raised a $13 million Series A, taking its total funding to more than $21 million. The round was led by Seligman Ventures, with participation from Sierra Ventures and 3one4 Capital.

Founded by Sudarshan Kamath and Akshat Mandloi, Smallest.ai develops the models and infrastructure behind enterprise voice agents. Its platform is used across financial services, healthcare, contact centers and business-process outsourcing.

The company is introducing Voice 4.0, an architecture designed to process listening, reasoning, actions and speech simultaneously. Traditional voice systems often pass a conversation through separate speech recognition, language-model and text-to-speech layers in sequence, creating delays and brittle interactions.

At the center of the platform is Hydra, a speech-to-speech model built to manage natural interruptions, use tools during calls and respond while a conversation is still developing. Smallest.ai also offers Pulse STT Pro, supporting 38 languages, and Lightning, its low-latency text-to-speech model.

The company works with organizations including RingCentral, Truecaller, Readymode, Piramal, Kogta and Pocket. It has nearly 60 employees and plans to expand its team as enterprise demand grows. Customer results and performance improvements remain company-reported.

The funding will support product development and wider deployment of enterprise voice agents across high-volume customer operations. The round also reflects growing investment in AI agent infrastructure and enterprise AI funding.

Smallest.ai’s challenge will be maintaining natural conversation quality while meeting the reliability, privacy and compliance requirements of large enterprise deployments.