StopDown
Smallest.ai logo
Funding · 24 January 2026Smallest.ai$8 MnVoice AI platform for enterprisessmallest.ai ↗

Smallest.ai launches voice AI platform with small models for faster, natural conversations

By · Editor ·

seedFounded 2023
  • Smallest.ai builds voice AI systems using small models to improve speed and responsiveness in real-time conversations
  • The startup launched in 2023 and has developed multiple versions of its speech stack, including text-to-speech, speech-to-text, and speech-to-speech pipelines
  • The company targets banking and financial services first, where voice is a more natural modality than text for customer interactions

What happened

  • Smallest.ai is a voice AI platform that uses small models to improve the speed and responsiveness of real-time conversations.
  • The startup was founded in 2023 by Akshat Mandloi and Sudarshan Kamath, who previously worked on AI systems in production for nearly a decade.
  • The company's focus is on building voice AI systems that are optimized for conversational use cases, rather than content creation.
  • Smallest.ai's technology is almost entirely proprietary, with models trained from scratch and tuned for specific architectural layers.
  • The startup has developed multiple versions of its speech stack, including text-to-speech, speech-to-text, and speech-to-speech pipelines.
  • The system supports emotional context, breathing patterns, and multilingual output across English, Spanish, major European languages, and top Indian languages.
  • Smallest.ai's technology is designed to be cloud-agnostic, allowing deployment on customer infrastructure for added security.
  • The company has secured clients in the banking and financial services sector in India and the US.

Why it matters

  • Voice AI is becoming a core interface for India's digital economy, which is expected to reach $1 trillion by 2030.
  • Smallest.ai addresses the gap in real-time conversational AI, where latency and responsiveness are critical.
  • The startup's focus on small models offers faster processing, lower costs, and reduced memory usage compared to large language models.
  • The company's approach to infrastructure allows deployment across different platforms, making it attractive to regulated enterprises.

The business

  • Smallest.ai provides a voice AI platform that includes text-to-speech, speech-to-text, and speech-to-speech pipelines.
  • The platform is designed for enterprise use, with support for real-time conversations and multilingual output.
  • The company's technology is optimized for conversational use cases, making it suitable for industries like banking and financial services.
  • Smallest.ai's models are trained from scratch and heavily tuned for specific architectural layers to improve speed and conversational flow.
  • The startup's system is cloud-agnostic, allowing deployment on customer infrastructure for added security.

The founders

Akshat Mandloi

co-founder
  • Mandloi has experience in building AI systems that run in production for nearly a decade.
  • He and Sudarshan Kamath are friends from college and co-founded Smallest.ai.
  • Mandloi believes that small models are more suitable for specific business use cases than large models.

Sudarshan Kamath

co-founder
in
  • He and Akshat Mandloi are friends from college and co-founded Smallest.ai.
  • Kamath has worked closely with Mandloi on developing the company's speech stack.

In numbers

$1 trillionIndia's digital economy · 2030
64msLatency for speech-to-text · 2026
sub 100msLatency for text-to-speech · 2026

What happens next

  • The company aims to build trust with regulated enterprises by offering flexible deployment options.

Up against

ElevenLabsCartesia

In their words

“Why do we need such large models to solve everything? To solve very specific business use cases, you don’t need very large models.”Akshat Mandloi, co-founder
“The final goal is how can we make conversations much more natural from a business perspective.”Akshat Mandloi, co-founder

Words to know

small models (SLMs)
AI models that are smaller in size and require fewer resources compared to large language models (LLMs), offering faster processing and lower costs.
cloud-agnostic
A system that is not tied to a specific cloud provider and can be deployed on different platforms.
real-time conversations
Interactions that happen immediately, with minimal delay, often used in voice AI for natural and responsive communication.
multilingual output
The ability of a system to support multiple languages in its output, such as text-to-speech or speech-to-text.

The numbers around it

Smallest.ai has 2 funding rounds in StopDown's archive, $21 Mn raised in all.

#The numbers around it
WhenRoundAmountLed by
1 August 2026Series A$13 MnSeligman Ventures

AI & Deep Tech in January 2026: 14 rounds with a disclosed amount, $193 million in all; the biggest was Emergent ($70M). Every sector this year →

Read it as a card in the app

Terms in this story

  • Startup: A startup is a young company built to find a new way of doing something and to grow fast, usually by serving many customers at little extra cost. India also has a legal definition for scheme benefits: under ten years old, turnover under ₹200 crore, and working on innovation or a scalable model.

Everything on Smallest.ai → · All startup news on 24 January 2026 →

Is this Smallest.ai? Get your “Featured on StopDown” badge →