Smallest.ai Raises $13M Series A to Break the Turing Test

3one4 Capital partners with technical founders possessing the operational rigour to challenge global software architecture. We first backed Sudarshan Kamath and Akshat Mandloi when Smallest was merely a research direction, and today we are proud to participate in their $13 million Series A round led by Seligman Ventures. Watching a focused team transition from a pre-seed stage thesis to an enterprise-grade platform powering millions of voice interactions within a year validates our conviction in their vertical-stack approach to applied artificial intelligence.

The Series A round brings the company's total funding to over $21 million. Seligman Ventures led the investment, with continued participation from Sierra Ventures and 3one4 Capital. The capital will accelerate the deployment of the world's fastest speech-to-speech operating layer for business communication.
‍

Solving the Latency Bottleneck in Real-Time Systems 

Human conversation is highly sensitive to timing. Natural dialogue depends on rapid, real-time responses. Conversational delays above 200 to 300 milliseconds feel robotic and unnatural to the human ear. Legacy voice automation systems fail to meet this standard because they rely on fragmented software pipelines. Traditional developers typically chain separate speech-to-text, large language models, and text-to-speech APIs together. Every transition between independent systems adds latency, causing cumulative delays that degrade the user experience.

Smallest solves this computational delay by collapsing the entire conversational loop into a single, unified execution plane. The company rejects the playbook of stitching third-party APIs together. Owning the full model stack allows the technical team to optimize performance across the entire pipeline. A millisecond of latency saved in transcription compounds directly into faster speech generation. The vertical integration guarantees an unbroken communication loop that remains stable under heavy enterprise workloads.

The Proprietary Model Stack Delivers Sub-Second Latency 

The technical team engineered a suite of specialized, compact models designed to run efficiently on production hardware. The vertical architecture consists of four interlinked technical layers:

• Lightning. The flagship text-to-speech synthesis model generates 10 seconds of highly expressive, lifelike audio in just 100 milliseconds. The model runs efficiently on less than 1 gigabyte of video random-access memory, allowing cost-effective edge deployment.
• Pulse. The specialized transcription engine hits first transcript in under 300 milliseconds. The speed matches human auditory recognition and ensures the voice agent begins processing user input immediately.
• Electron. The small language model designed for real-time dialogue outperforms general-purpose models on reasoning latency. The architecture decouples logical reasoning capability from raw parameter count to deliver fast, contextual responses.
• Hydra. The native speech-to-speech model represents the company's most advanced research milestone. The model processes incoming audio waveforms directly without converting them to text first. Asynchronous listening and thinking allow the agent to execute complex tool calls mid-sentence without breaking the natural flow of conversation.

The specialized models work in tandem within a unified orchestration layer. The integrated design allows the system to communicate with the natural cadence of a human operator, incorporating pauses, appropriate inflections, and emotional nuance.

Orchestrating High-Concurrency Workloads Under Enterprise Demands 

‍

Enterprise voice traffic is highly unpredictable and prone to sudden, concentrated surges. Standard contact centers experience sharp spikes in call volume during billing cycles, promotional events, or system disruptions. Traditional API-dependent voice agents frequently hit rate limits or experience severe performance degradation under high-concurrency conditions. Smallest mitigates this operational risk by controlling its own inference infrastructure.

The platform handles extreme traffic surges without relying on external servers. Operating independent infrastructure allows the company to manage load queues and provision compute resources dynamically. The platform supports on-premise deployment, delivering absolute data privacy for regulated industries. Smallest secures its enterprise-grade system with full compliance across major regulatory standards including HIPAA, GDPR, SOC 2 Type II, and ISO 27001.
‍

Enterprise Traction Proves the Architectural Moat 

Category-defining technology companies prove their moats through commercial deployment rather than temporary product demonstrations. Smallest has scaled from an early research project to a production-grade infrastructure platform. High-volume communications platforms like RingCentral and Truecaller use the Smallest architecture to power millions of customer interactions each month.

The technical efficiency of the stack also translates into commercial advantages. Smallest has reduced text-to-speech operational costs from 20 cents per minute to 1 cent per minute at scale. The cost reduction makes real-time voice automation economically viable even for thin-margin enterprises. 
‍

A Partnership Grounded in Engineering Excellence 

Our continued investment in Smallest reflects our belief in deep technical execution. We first partnered with Sudarshan & Akshat at the pre-product stage, and our conviction has deepened as the team continues to execute against their technical thesis as the market has evolved. The rapid progression from pre-seed validates the massive market demand for high-performance voice systems.

The future of applied artificial intelligence belongs to companies that build resilient, specialized systems designed for exact business outcomes. Smallest has shown that India origin teams can design, manufacture, and export foundational AI infrastructure to serve the world's most demanding enterprises. We are proud to continue our partnership as the team builds the definitive autonomous execution layer for consumer interaction.

‍

DISCLAIMER

_{The views expressed herein are those of the author as of the publication date and are subject to change without notice. Neither the author nor any of the entities under the 3one4 Capital Group have any obligation to update the content. This publications are for informational and educational purposes only and should not be construed as providing any advisory service (including financial, regulatory, or legal). It does not constitute an offer to sell or a solicitation to buy any securities or related financial instruments in any jurisdiction. Readers should perform their own due diligence and consult with relevant advisors before taking any decisions. Any reliance on the information herein is at the reader's own risk, and 3one4 Capital Group assumes no liability for any such reliance.Certain information is based on third-party sources believed to be reliable, but neither the author nor 3one4 Capital Group guarantees its accuracy, recency or completeness. There has been no independent verification of such information or the assumptions on which such information is based, unless expressly mentioned otherwise. References to specific companies, securities, or investment strategies are not endorsements. Unauthorized reproduction, distribution, or use of this document, in whole or in part, is prohibited without prior written consent from the author and/or the 3one4 Capital Group.}

‍

‍