How Vobiz.ai Is Powering the Next Generation

Before founding Vobiz.ai, Gandham had already built and exited a startup. His consumer neobank Finin was acquired by Open, after which he took a sabbatical before returning to build AI voice agents for financial services use cases.

It was during that work that the infrastructure problem became impossible to ignore. He partnered with Srivastava, who had spent years inside telecom infrastructure companies including Plivo and Bandwidth, and together they began rebuilding the telecom layer specifically for AI-native communication systems.

The core issue is deceptively simple. Traditional telecom infrastructure introduces 300 to 500 milliseconds of delay on its own. When you stack speech-to-text processing, large language model inference, and text-to-speech generation on top of that, total response time can easily exceed 1.5 seconds.

At that threshold, the experience breaks down.

"At that point, humans immediately recognise they are speaking with AI," Srivastava explains.

There is also a reliability problem that does not exist in human calls. In a normal conversation, if audio cuts for two seconds, you simply ask the other person to repeat. In an AI conversation, the same interruption can cause the model to misinterpret the context entirely and respond with something completely off-track.

These are not edge cases. They are structural limitations of infrastructure that was never designed with AI in mind.

A typical voice AI stack has four layers: the large language model, speech-to-text, text-to-speech, and telephony infrastructure. Vobiz.ai focuses entirely on that last layer, the one closest to the actual phone call.

The company has built what it calls a single-hop architecture designed to reduce delays, background noise, and latency across the entire call. It integrates with multiple AI orchestration and speech providers including OpenAI, Gemini, ElevenLabs, Cartesia, AWS Polly, and LiveKit, dynamically routing workloads across providers depending on performance, latency, language requirements, and use case suitability.

The platform also applies AI internally across its own infrastructure for real-time media optimisation, echo cancellation, noise suppression, packet routing, spam detection, answering-machine detection, call streaming, and transcription support.

The result, according to the founders, is telephony latency reduced to under 80 milliseconds at P95 levels: meaning 95% of calls experience that delay or less.

Developer experience has been another focus. Instead of the lengthy provisioning cycles that characterise legacy telecom providers, Vobiz.ai customers can complete KYC, provision numbers, access APIs, and deploy integrations through a self-serve onboarding flow in minutes.

Since launching in November last year, the platform has scaled from roughly 1 lakh calls per month to more than 10 lakh calls per day. Customer retention sits at 98%.

Vobiz.ai's first wave of customers came from India's fast-growing voice AI startup ecosystem. Its client roster includes Bolna, Sarvam AI, Razorpay, RevRag, Smallest.ai, and Navana AI, among others.

Early on, nearly all business came from startups building conversational agents for enterprise clients. That mix is now shifting. Enterprises are increasingly buying directly, particularly across fintech, lending, insurtech, logistics, and real estate.

Today enterprises account for roughly 30% of Vobiz.ai's business, with AI voice startups contributing the remaining 70%. The founders expect enterprise adoption to overtake startup-driven demand by the end of the year as more companies move from experimentation into production deployments.

The economics reflect an infrastructure business rather than an application-layer AI startup. Gross margins are expected to stay in the 50 to 80% range depending on usage patterns and contract structures. Revenue comes from three streams: telecom numbers, usage-based call billing, and value-added services such as transcription, call streaming, and answering-machine detection. The founders expect value-added services to become the largest revenue contributor over time.

Vobiz.ai positions itself alongside global players like Twilio and Telnyx and homegrown competitors like Exotel and Plivo. But unlike those incumbents, which were built for human communication, Vobiz.ai is designed from the ground up for AI-native workloads.

While voice remains the primary focus today, Vobiz.ai is already expanding into adjacent channels. WhatsApp calling and chat integrations are live, with SMS and RCS infrastructure planned next. The broader ambition is to become a default AI communication layer across all channels, not just voice calls.

Geographically, the company is preparing for international expansion across the US, Europe, the Middle East, Africa, and Asia-Pacific, with an initial focus on markets where AI telecom infrastructure remains underpenetrated rather than competing head-on with incumbents in saturated Western markets.

Compliance has been built into the platform from the start, covering DPDP, GDPR, SOC2, and ISO standards.

"We built the platform as an international product from day one," Gandham says.