Analysis · AI & Deep Tech

Sarvam AI co-founders share vision and progress

By Startup Enthusiast · 17 March 2026

What they do

Sarvam AI builds India's full-stack sovereign AI platform with voice and translation tools

Co-founders Pratyush Kumar and Vivek Raghavan gave an interview on 17 March 2026

They have spent $25–30 million and are building smart glasses called Sarvam Kaze

Sarvam AI is India's full-stack sovereign AI platform.
Co-founders Pratyush Kumar and Vivek Raghavan gave an interview on 17 March 2026.
The platform includes Voice Agents, Text to Speech, Speech to Text, Translation, and Document Digitisation.
Voice agents handle use cases like Cart Recovery Nudge, EMI Payment Reminder, and Appointment Booking.
Text to Speech supports 11 Indic languages; Speech to Text supports 12 Indic languages; Translation works across 23 languages.
The company has spent $25–30 million in total funding.
It received support from the India AI Mission — access to compute at concessional rates with an equity-based structure.
The team is building smart glasses (Sarvam Kaze) to make intelligence accessible anytime, anywhere.

Co-founders believe India should build its own AI laboratory and foundational technology to avoid becoming a digital colony.
They cite Aadhaar and UPI as templates for AI adoption in India.
Sam Altman (OpenAI CEO) previously suggested India should not attempt building foundational AI models.
The Voice AI platform handles millions of minutes of voice calling per day, growing double-digit percentages per month.

Sarvam AI is a full-stack sovereign AI platform with Voice Agents, Text to Speech, Speech to Text, Translation, and Document Digitisation.
Voice agents handle use cases like Cart Recovery Nudge, EMI Payment Reminder, and Appointment Booking.
Text to Speech supports 11 Indic languages; Speech to Text supports 12 Indic languages; Translation across 23 languages.
Offers REST API, Python SDK (pip install sarvamai), and a browser playground.
Deployment options: Sarvam Cloud, Private Cloud (VPC), On-Premises (air-gapped).
Security: SOC 2 Type II, ISO 27001, DPDP compliant, role-based access, audit trails, data residency controls.
Tokens served: 10B+; median latency <100ms; uptime SLA 99.9%.
Voice AI platform handles millions of minutes of voice calling per day, growing double-digit percentages per month.

Primary focus over next several months: reach — making technology accessible to more people and building infrastructure for availability and access.
Building smart glasses (Sarvam Kaze) to make intelligence accessible anytime, anywhere.

Product: Foundation models and agentic AI systems with voice capabilities and enterprise applications
Customers: Enterprise customers across finance, aviation, telecom and government sectors
How it makes money: SaaS and licensing model for AI models and enterprise AI applications
What sets it apart: India-built sovereign AI stack with focus on cost-effective deployment and localization