While current AI agents can solve problems, they often sound like machines. Smallest.ai aims to change that by creating voice models that mimic human conversation. The startup's approach is not about making larger language models faster but about building smaller, more specialized ones designed to process and respond in real-time with minimal lag.

Founder Sudarshan Kamath believes the future of AI agents lies in coupling small, highly responsive voice models with large foundational models for complex queries. These voice models can handle diverse accents, multiple languages and noisy environments – making them ideal for customer support applications where natural conversation is crucial.

The startup has secured $13 million to develop its technology, bringing its total funding to over $21 million. This investment will help Smallest.ai grow its existing customer base, which includes companies like RingCentral and Truecaller, while also expanding into new markets.

With competitors like ElevenLabs already in the field, Smallest.ai is pushing the boundaries of what voice AI can do, aiming to make their models indistinguishable from human conversations. The ultimate goal? To pass the Turing test for AI voices and revolutionize how businesses interact with customers.