Gnani

Gnani develops proprietary speech and language models for enterprise applications as a frontier Voice AI company, building foundational models trained on 14 million hours of real telephonic audio across 40+ languages. Its model stack includes Gnani Prisma v2.5 for speech-to-text (ranked #1 in 8 of 9 Indian languages on the Kathbath Noisy benchmark), Gnani Timbre v2.0 for text-to-speech in 21+ languages, Gnani Warp v2.0 as India first 5-billion-parameter speech-to-speech model, and Gnani Aion v3.2 language models scoring 37.99% on Berkeley BFCL v3 for agentic function calling, outperforming GPT-4o-mini and Claude Sonnet 4. Enterprise products built on these models include voice agents for collections, customer service, KYC, and outbound calling; omnichannel analytics and quality assurance; AI-powered voice biometrics for authentication; and Agent Assist providing sub-500ms guidance during live calls. The platform processes over 30 million daily voice interactions across 200+ enterprise deployments in banking, insurance, healthcare, government, and telecom. Gnani supports cloud, private cloud, on-premise, and air-gapped deployments with SOC 2, ISO 27001, GDPR, HIPAA, and PCI-DSS compliance. Founded in 2016 and headquartered in Bengaluru, India, Gnani has raised approximately $21.9 million in Series B funding from investors including Samsung Venture Investment, Info Edge Ventures, and Aavishkaar Capital.