Maxim
Serebryakov

A Friend's Humiliation. A $149M Answer.

There's a moment that changes everything. For Maxim Serebryakov, it happened in a contact center somewhere in the orbit of a customer service script and a manager's ultimatum. A friend - Maxim would later call him "Raul" in interviews - was told to either neutralize his accent or lose the job. Not a performance issue. Not a skill gap. Just the way his voice landed on the other end of the line.

Serebryakov grew up between New York and Russia, arriving in the U.S. for college with the particular ear that comes from navigating languages and borders from childhood. He understood what was actually being asked of Raul: not clarity, but conformity. The solution he built with two Stanford classmates doesn't ask people to change anything about themselves. It changes the signal, not the speaker.

Sanas - named from the Spanish word meaning "to heal" - launched in 2020. Three international students from Russia, China, and Venezuela sat at Stanford and decided they'd rather fix the technology than the human. The company's AI runs on the agent's own machine, in under 150 milliseconds, modulating accent while preserving what matters: voice, warmth, emotion, personality.

150 Milliseconds to Better Understanding

Sanas built what it calls the world's first Real-Time Speech Understanding Platform. The core insight is architectural: rather than sending audio to the cloud and waiting for a processed response, the entire model runs locally on the agent's computer. Privacy by design, not by policy.

The latency - 150ms - is worth dwelling on. The human brain's own audio processing lag runs around 200ms. Sanas completes its transformation faster than consciousness catches up. In a live customer service call, the gap is imperceptible. The agent speaks, the customer hears clarity. Neither party stops to notice.

The numbers are enterprise-grade. An 18% reduction in average handle time in a contact center with thousands of agents translates directly to eight-figure cost savings. The 22% CSAT improvement means fewer escalations, fewer callbacks, fewer churned customers. This is why Teleperformance - one of the world's largest BPO operators - didn't just invest in Sanas's Series B. They became a strategic partner.

The Team That Built It

There's a version of this story that's pure Stanford mythology: three foreign-born students in a university lab solving the problem they personally lived. The reality is just as clean. Each co-founder brought a different piece of the puzzle from a different part of the world.

It's worth noting that "Sanas" - the word Andrés contributed from Spanish - carries both the act of healing and the plural of "sana" (healthy). The name is doing double duty. The company is healing accent bias while making communication healthier. A small, careful choice from a team that would go on to be precise about everything.

From Seed to $149M: Building on Conviction

The Series B in February 2025 told a specific story about where enterprise AI money is flowing. Quadrille Capital led. Insight Partners returned. But the detail worth noting is Teleperformance - the global BPO giant with 300,000+ employees - writing a check alongside Alorica, one of the largest North American contact center operators. When your largest potential customers are also investors, you've crossed a threshold.

Serebryakov used the language of infrastructure, not luxury, when announcing the round: "Accent-based bias is a global challenge. At Sanas, we're building toward a future where people can more easily understand each other, regardless of their origin." The language of infrastructure suggests he sees Sanas as table stakes for global communication, not a premium add-on.

From Stanford Lab to Global Platform

Accent Modulation as Infrastructure

Serebryakov makes a specific comparison in interviews: he wants accent translation to become as standard as noise cancellation. That's a loaded analogy. Noise cancellation was a niche audio engineering feature in 2000; by 2020 it was a baseline expectation on every video call. The technology dissolved into infrastructure. Nobody thinks about it. Everyone benefits.

His expansion roadmap tracks that ambition. The Series B funds three distinct vectors: new speech-to-speech algorithms, new geographies, and new industries beyond contact centers. The contact center market is large - tens of millions of agents globally - but it's a beachhead, not a destination. Anywhere voice crosses a border is Sanas territory.

There's also a harder argument embedded in the product. Running locally means Sanas processes no audio in the cloud - a deliberate choice in an enterprise market where data sovereignty is non-negotiable. The privacy architecture isn't a marketing claim; it's why companies in financial services and healthcare are deploying it at scale. ISO compliance and GDPR-readiness aren't afterthoughts; they're the product.

The Immigrant's Frequency

Something worth acknowledging: Serebryakov built an accent-modulation company as someone who navigated multiple languages and cultures growing up. Born in New York, raised in Russia, returned to the United States for Stanford. He knows both sides of the call center dynamic. He's been the person whose accent marked him as "other" in a room. The product isn't abstract empathy engineering. It's a specific solution to a specific wound.

His two co-founders bring the same texture. Shawn Zhang built systems at Cisco before co-founding Sanas - he knows what enterprise-grade latency tolerance actually looks like. Andrés Pérez Soderi, who contributed both the company name and the Venezuelan perspective on what accent discrimination costs in professional contexts, holds the financial and operational architecture. Three problems the world handed them. One company to fix it.

At 310 employees and $33.2M in annual revenue, Sanas is no longer a Stanford experiment. It's operating infrastructure for hundreds of thousands of people who need to be heard clearly every single shift. Maxim Serebryakov is still running it with the urgency of someone who knows exactly why it matters.