Sanas Real-Time Speech AI Platform

Sanas.AI Inc. is a speech AI company founded in 2020 at Stanford University and based in Palo Alto, California. It targets a core problem of offshore and multilingual contact centers: customers and agents misunderstanding each other because of accents, languages and background noise. The website reports more than 750,000 users globally.

The platform processes speech in real time on live calls. Accent Translation converts an agent's accent while keeping the speaker's voice and emotion, Language Translation translates speech across 25+ languages, and Speech Enhancement removes noise, echo and reverb. Speech Intelligence and Supervised AI add real-time insight and oversight, and a developer SDK lets other products embed the models. It works alongside contact center platforms such as Genesys, NICE, Five9, Talkdesk and Zoom, and serves BPOs, healthcare, financial services, retail, travel and telecom companies.

• Accent Translation: converts accents in real time while preserving voice identity and emotion.
• Language Translation: real-time speech translation across 25+ languages.
• Speech Enhancement: removes background noise, echo and reverb on both sides of the call.
• Speech Intelligence: real-time analysis of live conversations.
• Supervised AI: oversight layer for AI and human voice interactions.
• Developer Platform: SDK for embedding Sanas speech models into other products.

Vietnamese BPO and IT outsourcing firms serving US, Australian or Japanese clients can make agents easier to understand on English-language support lines. Contact centers running from noisy shared floors or remote setups can clean up call audio. Software vendors can embed the SDK for real-time translation in voice products.

Agents with different accents can be understood clearly by customers in other markets without retraining. Noise cancellation lets agents work from busy floors or home. The website states that call audio is not monitored, recorded or stored, and lists HITRUST, HIPAA, SOC 2, ISO 27001, GDPR and PCI DSS compliance, which matters for banking and healthcare clients.

Asia hosts a large share of global BPO and offshore contact center work, and buyers increasingly expect clear, natural voice service across accents and languages. Real-time speech-to-speech AI that sits in the call path is emerging as an alternative to accent training and as a way to serve customers in their own language.

Founded in 2020 at Stanford by Shawn Zhang, Max Serebryakov and Andres Perez Soderi, with CEO Sharath Keshava Narayana joining in 2021. Raised a 5 million USD seed, a 32 million USD Series A led by Insight Partners and a 65 million USD Series B in 2025. Holds US and India patents for real-time accent conversion and partners with Mavenir for in-network voice AI.