Michael Ovitz Backs Agentic AI Startup Vapi With $20M to Build Voice Agents

Back in 2023, Jordan Dearsley was navigating the complexities of voice AI technology while running an AI-powered voice therapist startup. It was his second go at entrepreneurship, having built a $500K ARR note taking app. But spending hours on daily walks with his therapeutic AI creation, Dearsley wasn’t seeking companionship; he was refining the infrastructure that would eventually evolve into Vapi. Recognizing the challenges developers faced when building voice AI solutions, he along with his co-founder, Nikhil Gupta, pivoted to a broader vision, creating a developer-first platform to simplify the creation of voice AI Agents.

Launched just nine months ago, Vapi lets developers build, test, and deploy voice agents in minutes rather than months, offering a streamlined path to production for a variety of applications. The startup recently secured $20 million in Series A funding, led by Bessemer Venture Partners, with participation from business magnate, Michael Ovitz, bringing its valuation to $130 million. From inbound and outbound calls to voice-enabled IoT and customer support solutions, Vapi’s versatile API empowers businesses to deploy voice technology with unprecedented ease.

Creating a voice AI agent with Vapi is a streamlined process for developers. The platform enables users to design conversation flows, configure voice pipelines, and deploy agents to handle tasks ranging from customer support to outbound sales. Vapi’s server SDKs, available in Python, Node.js, and other languages, further simplify integration into applications.

In a market crowded with Agentic AI startups like Decagon and Sierra, the agentic AI builder space is currently one of the hottest areas in tech. Frameworks like LangChain and CrewAI cater to developers with technical expertise, while AI Agent builder platforms like Langbase are simplifying the process further. Vapi carves out its own niche by focusing exclusively on voice AI agents, addressing industries where voice is a critical operational component, such as healthcare and finance. "We're targeting companies where voice is the primary channel for revenue," explained Dearsley in an interview with StartupHub.ai. "For some companies, particularly those with an older demographic, such as healthcare, finance, and insurance, voice is the primary interface. That's our target customer."

While giants like OpenAI and Google pose a threat to move downstream from their foundation model development, Vapi distinguishes itself with its focus on specialized enterprise needs. “Large companies take a generalized approach, but they’re not addressing niche use cases like HIPAA-compliant real-time APIs for healthcare,” explains Dearsley. “Our platform is built for these last-mile production problems, making us indispensable to enterprises. The incumbents are going to be flat-footed here.”

The platform’s ability to tackle real-time interaction challenges is another standout feature. "It's very interesting how model architectures have evolved over the past year alone," Dearsley shared. "We've seen latency and cost improvements, better and closer performance, and iterations of architectures from multimodal to unified models that can hear frustration in a customer's voice and produce an empathetic response. For open-source models like Moshi by Kyutai, they can hear and produce in real time, even affirming other people's speech like 'yup, got it,' and overlapping in conversation. The average latency is 320 milliseconds, and humans actually overlap each other in conversations. To reach human performance, you need speaking and hearing simultaneously. Our goal is to get voice agents ready to speak in real-time, and when a new model comes out, our tech can refer the best model appropriate while letting our users specify their preferences."

Vapi also considers context windows critical to ensuring reliability and determinism in voice agents. In fact, they want to restrict the context window length. "Our goal is determinism," explained Dearsley. "We can make them reliable in terms of uptime, but not behavior. We want to limit the context, like collect first name, collect last name, then share test results. We need a system that can inject the right context at the right moment."

Bessemer, their lead investor, has penned their thesis on the sector as of late, highlighting Voice AI as a transformative force poised to redefine customer interactions across industries. Their recent analysis shows how voice technology is moving beyond simple commands to deliver end-to-end services. Traditional IVR systems, which still represent a $5 billion market, fail to meet the demands of modern, fluid conversations. Emerging voice-native models, capable of processing raw audio directly, promise faster, more natural interactions with latency as low as 300 milliseconds. Bessemer believes execution speed and natural conversational quality will define the next wave of innovation, making platforms like Vapi critical to the market’s evolution.

Vapi’s platform currently supports over 100,000 developers and powers hundreds of startups, "ripping millions of API calls weekly," noted Dearsley. Customers like Mindtickle, Luma Health, Ellipsis Health, and Gestionadora de Créditos rely on Vapi to handle custom voice agent deployments. “Vapi’s mature platform enables seamless integration, helping us deliver solutions to our customers in record time,” says Marcelo Oliveira, SVP of Engineering at Luma Health.

With its recent funding, the startup aims to expand its workforce and enhance its core infrastructure, ensuring it remains at the forefront of voice AI innovation.

“Apple Intelligence and Google Gemini are poised to onboard billions of people to voice assistants that truly converse like humans,” added Dearsley. “Enterprises need a robust platform to meet this demand, and we’re here to provide that foundation.”

The funding round also saw participation from Abstract Ventures, AI Grant, Y Combinator, and Saga Ventures.