Phonic is a San Francisco-based voice AI company founded by MIT graduates Moin Nadeem and Nikhil Murthy. The company is built around a contrarian technical bet: most voice agents today are assembled from a pipeline of third-party components — a speech recognition vendor, a large language model, and a text-to-speech provider — each adding latency, cost, and points of failure. Phonic instead trains a unified speech-to-speech model in-house, end to end, so the system understands and generates voice in a single tightly integrated stack.
This architecture has two practical payoffs. First, latency drops, because audio does not have to round-trip through several external APIs, which makes conversations feel responsive and human rather than stilted. Second, because Phonic owns and hosts its own models, it can run them far more cost-efficiently than a stack reliant on per-call charges from multiple vendors — an advantage that compounds for customers running millions of minutes.
Phonic emphasizes reliability as a first-class feature, providing tooling to build, evaluate, and monitor voice agents so they behave consistently in production. The platform is aimed at developers and enterprises that need conversational voice for support, scheduling, intake, and other phone-based workflows.
The company raised a $4M seed round led by Lux Capital, with participation from a notable roster of operator-angels including Replit co-founder Amjad Masad, Hugging Face co-founder Clem Delangue, Applied Intuition co-founder Qasar Younis, and Modal Labs founder Erik Bernhardsson. Lux partner Grace Isford highlighted Phonic's in-house, end-to-end training approach as the key differentiator. Phonic has subsequently expanded its backing with later funding from firms including First Round Capital and Felicis, reflecting growing investor conviction in vertically integrated voice AI.