Smallest.

ai, a startup founded in late 2024, has raised $13 million in a Series A round led by Seligman Ventures, with participation from Sierra Ventures and 3one4 Capital, bringing total funding to over $21 million.

The company is building a small, specialized voice AI model designed to make real-time conversation with an AI agent indistinguishable from speaking to a human.

Unlike large language models that process entire prompts before responding, Smallest.

ai’s model mimics human conversation by listening, thinking, and speaking simultaneously, enabling virtually zero response lag.

The startup’s existing customers include RingCentral and Truecaller, and it targets enterprise customer support use cases.

When the model encounters a topic outside its limited knowledge base, it hands off the query to a larger foundational model, briefly placing the customer on hold — similar to how a human would pause to research an issue.

Founder and CEO Sudarshan Kamath believes future AI agents will rely on two models: a small voice model for real-time interaction and an offline LLM called on for complex problems. Smallest.

ai focuses strictly on voice-specific nuances like handling diverse accents, supporting dozens of languages, and operating in noisy environments.

It competes with ElevenLabs, Cartesia, and regional players like Sarvam, but deliberately avoids use cases like audio dubbing or podcasting, instead concentrating on real-time conversational voice agents.

Kamath emphasized that the company’s sole focus is to break the Turing test — making users unable to tell whether they are speaking to an AI or a human.

The article also notes that for customer support startups, building voice AI internally can be a distraction from their core business, positioning Smallest.ai as a specialized layer.