The human computing company — real-time conversational video avatars via Phoenix models
HQ: San Francisco, CA, United States | Founded: 2021 | Employees: 40–80 (est. 2026) | Stage: Series B ($40M) | Website: https://www.tavus.io
Tavus (Tavus Inc.) was founded in 2021 in San Francisco by Hassaan Raza (CEO) and Quinn Favret (COO/CPO), who met experimenting with personalized video to sell software — then left the sales-tool path after raising from Sequoia to return to AI research. It went through Y Combinator W21. The company develops Phoenix — a family of real-time conversational video models — and a replica platform where users clone their face/voice to generate personalized videos and live avatar conversations via API. It raised $6.1M seed led by Sequoia (2021), $18M Series A led by Scale Venture Partners in March 2024, and $40M Series B led by CRV in 2025. In late 2024-25 CEO Hassaan Raza publicly detailed pivoting back to research roots despite paying customers, framing Tavus as a 'human computing' research lab.
Tavus won by going research-first after nearly going sales-tool: Hassaan Raza and Quinn Favret discovered personalized video (angry at a sales job) then walked away from paying customers after Sequoia money to build Phoenix — its real-time avatar models — positioning as 'human computing' not video generation. YC W21 → $6.1M Sequoia seed → $18M from Scale Venture Partners (Mar 2024) → $40M Series B led by CRV (2025). Real-time responsiveness (lip-sync, turn-taking) is the wedge vs HeyGen, Synthesia and D-ID; but frontier labs and ElevenLabs compete for the same real-time human face/voice real estate.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki