San Francisco-based Tavus announced a $40 million Series B round led by CRV, with participation from Scale Venture Partners, Sequoia Capital, Y Combinator, HubSpot Ventures and Flex Capital. The capital will fund the scaling of its research team, broaden deployment of its Personal Affective Links (PALs), and deepen the development of the proprietary models that power them.
Founded by CEO Hassaan Raza, Tavus is positioning PALs as the next evolution of human-computer interaction: AI agents that see, hear, understand emotion and act with the nuance of a human counterpart. The round underscores investor confidence in Tavus’s vision of “human computing,” a market where AI moves beyond text-only chatbots to truly multimodal, emotionally intelligent companions.
Series B Signals Strong Investor Backing
The $40 million raise brings together a roster of seasoned backers that have previously backed leading AI and SaaS innovators. CRV led the round, while Scale Venture Partners, Sequoia Capital, Y Combinator, HubSpot Ventures and Flex Capital joined as co-investors, reflecting broad consensus on Tavus’s strategic direction.
The funds are earmarked for expanding the company’s research talent, scaling the PAL platform for enterprise adoption, and accelerating the rollout of new model iterations that enhance realism, latency and contextual awareness.
Personal Affective Links: AI Humans With Agency
Tavus’s PALs are described as “emotionally intelligent, multimodal, and capable of understanding and perceiving” digital agents that interact via video, voice and text. Each PAL maintains a lifelike visual presence, can read facial expressions, gestures and tone, and remembers context to act autonomously, scheduling meetings, sending messages, or completing tasks without explicit prompts.
Raza emphasizes that PALs “meet users where they are,” evolving with each interaction to handle increasingly complex scenarios across recruiting, sales, education and customer service.
Meet the models: Phoenix-4, Sparrow-1, Raven-1
Behind every PAL sits an in-house suite of foundational models:
- Phoenix-4 – a state-of-the-art rendering engine that drives lifelike facial expression, head-pose control and emotion generation at conversational latency.
- Sparrow-1 – an audio-understanding model that blends deep conversational intelligence with semantic emotional cues, managing timing, tone and intent in real time.
- Raven-1 – a contextual perception system that interprets people, environments, gestures and emotions, giving PALs a genuine sense of presence.
These models are orchestrated by a proprietary memory-management layer, enabling seamless multimodal interaction that feels “second nature,” according to Raza.
Market Traction and Ecosystem Adoption
Tavus reports that more than 100,000 developers and enterprises already leverage its platform across a range of verticals, from talent acquisition to education. Early PAL deployments have demonstrated the technology’s ability to improve engagement metrics and reduce friction in human-AI workflows. The company continues to refine its models and expand client pilots, using real-world interaction data to fine-tune responsiveness and consistency.
Why Zillionize Invested
Zillionize invested in Tavus during Y Combinator’s Summer 2021 batch because we recognized the transformative potential of AI-powered video communication and the founding team’s vision for making digital interactions more personal and authentic. Since our initial investment, Tavus has evolved from a personalized video platform into a pioneering AI research company, developing proprietary foundational models that enable machines to see, hear, understand, and interact with people in remarkably human ways. The introduction of PALs (Personal Affective Links) represents a significant leap forward, combining emotional intelligence, contextual awareness, and autonomous capabilities to redefine how people interact with technology. With more than 100,000 developers and enterprises using its technology, Tavus is demonstrating both the technical innovation and market traction needed to help establish human computing as the next major evolution in artificial intelligence. We’re excited to continue supporting the team as they work to make human-like AI interaction a fundamental part of everyday computing.