The developer platform for voice AI — $70M raised, ~$500M valuation after powering Amazon Ring's calls
HQ: San Francisco, CA, United States | Founded: 2023 | Employees: ~50-100 | Stage: Series B ($50M at ~$500M valuation, May 2026); $70M total raised | Website: https://vapi.ai
Vapi is a developer platform for AI voice agents. One API gives builders the full stack: speech recognition, LLM orchestration, speech synthesis, and telephony — orchestrated for sub-second, natural phone conversations with tool-calling into business systems. Developers configure agents in code or dashboard, test them with built-in tooling, and scale them across inbound and outbound calls with real-time observability.
The company was founded in 2023 by Jordan Dearsley and Nikhil Gupta, who had both felt the pain of building voice AI the hard way — stitching models, handling latency, debugging telephony. They went through Y Combinator (W24), launched the platform, and reached millions of dollars in revenue within roughly six months. In December 2024 they raised a $20M Series A led by Bessemer Venture Partners at around $130M valuation, with participation from Abstract, AI Grant, Y Combinator, Saga Ventures and Michael Ovitz.
The breakout came in May 2026: a $50M Series B led by Peak XV Partners at roughly $500M valuation, announced after Vapi's infrastructure was revealed to be handling 100% of Amazon Ring's customer-service calls — production-scale proof that its platform could carry real consumer volume reliably. The round funds engineering expansion, enterprise tooling, and international growth.
Vapi positions itself as model-agnostic infrastructure: swap in OpenAI, Gemini, Anthropic or open-weight models as they improve, keep the runtime, integrations and observability layer constant. Its customers range from startups building AI sales agents to enterprises automating support and scheduling.
Vapi is building the missing infrastructure layer for voice AI: the equivalent of what Twilio did for SMS and Stripe for payments. Building a phone agent that doesn't feel broken — sub-second responses, clean interruption handling, reliable tool-calls into real systems — requires gluing together speech recognition, LLMs, synthesis, and telephony while managing latency at every hop. Vapi sells that glue as one API, so developers ship agents in days instead of quarters.
The traction is real, and the proof point is a monster: in May 2026 Vapi raised a $50M Series B led by Peak XV at roughly $500M valuation on the news that its platform was handling 100% of Amazon Ring's customer-service calls — a production deployment at genuine consumer scale, not a pilot. That came on top of a $20M Series A (Bessemer, Dec 2024, with Abstract, AI Grant, YC, Saga Ventures and Michael Ovitz), taken when the company had already reached millions in revenue within its first six months — at just $130M valuation, leaving plenty of room for the 4x step-up.
The bet underneath: voice is about to be the interface for most business transactions, and the models alone won't get you there — someone has to own the runtime, the telephony, the integrations, the observability. The risks are the voice-infrastructure standard ones: model labs shipping ever-better native speech-to-speech, rivals Bland (more enterprise/self-host) and Retell (more efficient), and pricing pressure as voices commoditize. But developer mindshare compounds, and Vapi currently owns it in the category's developer tier.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki