Seattle-based Nuance has raised a $50 million Series A round to build AI models that can hold face-to-face conversations with less lag and fewer awkward moments than current chatbots.
Nuance says the company is trying to make AI better at conversation and emotional intelligence. Fangchang Ma, the cofounder and CEO, said existing voicebots and AI avatars are often built by connecting voice-to-text, a large language model, and text-to-voice systems. He said that setup creates latency and misses human nuances like reactions and microexpressions.
Ma and his two cofounders all worked at Apple before starting Nuance. He said the startup chose a harder path by building 'one system, on one model' with 'audio video in and audio video out.'
The company is still pre-product, but Ma said he sees several uses for it. Those include video calls with an AI to practice interviews, give presentations, or learn a new language. Nuance also plans to sell to businesses, with possible use cases that include AI interviewers, sales reps, and customer support.
Nuance said the model cannot be trained by scraping the internet. Instead, it collects data in-house or from data partners. Ma said the ideal training data would capture both sides of a real conversation, with clean audio and natural, non-performative interaction.
The startup plans to open a research preview of its model to the public later this year.
The round was led by Lightspeed Venture Partners, with participation from Accel, Nvidia's NVentures, South Park Commons, and Define Ventures. It follows a $10 million seed round led by Accel in July last year.
Ma said Nuance has used the funding to grow its eight-person team. The company is hiring for research and development and plans to bring on go-to-market experts to sharpen pricing and monetization.
For founders, the signal is familiar: capital is still flowing to AI, but investors want a clear technical edge and a path to revenue. Nuance is pitching both — a more unified model architecture and a business plan that goes beyond demos. Founders who get this right move faster, because the product is not just smarter, it is easier to sell.
