Official information about Tavus

Basic information

Name: Tavus
Type: Human Computing / Foundational AI Research Lab
Founded: 2020 (Y Combinator, Summer 2021 batch)
Headquarters: San Francisco, California, United States
Founders: Hassaan Raza (CEO) and Quinn Favret (COO)
Website: tavus.io
Category: Human Computing (full-stack AI humans, called PALs, and the foundational models that power them)
Funding: $40M Series B led by CRV (November 2025), with participation from Scale Venture Partners, Sequoia Capital, Y Combinator, HubSpot Ventures, and Flex Capital. Approximately $64M raised to date.

Background

Tavus was founded in 2020 by Hassaan Raza and Quinn Favret to solve a fundamental gap. Automation has scaled efficiency but stripped away empathy, nuance, and presence from digital interactions. Tavus pioneered a new category called Human Computing, teaching machines the art of being human so that talking to a computer feels as natural as talking to a friend or coworker. Its foundational models teach machines to see, hear, understand, remember, and respond with emotional intelligence in real time, and its PALs bring those models to life as full-stack counterparts people can talk to face to face.

Today, Tavus is the leader in human simulation models, differentiating sharply from asynchronous video generation tools. More than 100,000 developers and enterprises including Amazon, Deloitte, EY, Mayo Clinic, CVS, Salesforce, Aetna, and Wix build on Tavus. Its mission is to close the gap between humans and machines by removing the friction of today’s systems.

Core services

Conversational Video Interface (CVI): The developer API for embedding real-time, face-to-face conversation into any product. CVI unifies perception, speech, language, rendering, and orchestration behind a single interface and supports bring-your-own LLM, external knowledge integration with retrieval, persistent memory, real-time guardrails, and function calling.
PAL Maker: The no-code builder for creating and deploying PALs. Users describe an idea to Charlie, the built-in guide, give their PAL a face, a voice, knowledge, and capabilities, and deploy it to a site or app in a few clicks.
PALs: Out-of-the-box counterparts that move across chat, voice, and face-to-face video. A PAL defines behavior, knowledge, personality, memory, and pipeline configuration, from system prompt through objectives, guardrails, and tools.
Faces: Photorealistic likenesses rendered by Phoenix-4, built from a single image or two minutes of video, with full-face rendering, natural lip sync, and identity preservation. A library of stock faces is available for immediate use.
Solutions: Enterprise deployments built, tuned, and managed by the Tavus team for sales, healthcare, learning and development, and recruiting, typically live within weeks.

Audience and use cases

Product teams and developers: embed Tavus CVI to deliver richer, humanlike product experiences without ballooning headcount.
Enterprise teams: partner with Tavus Solutions for onboarding, training, customer support, and coaching without heavy engineering lift.
Individuals and prosumers: create PALs for personal productivity, tutoring, or on-demand services.

Key use cases include:

AI recruiters and interviewers
Role-play for education and sales training
Healthcare intake and patient navigation
Customer onboarding and support portals
AI tutors and coaches
Companionship and wellness support
Sales outreach and personalized video

Clients and case studies

ACTO (life sciences): replaced role-play training with Tavus-powered practice partners, boosting pharma rep confidence and scalability.
Orum (sales enablement): embedded Tavus-powered role-play so reps practice cold calls, discovery, and objection handling on demand, with scenarios tuned by buyer type, difficulty, and sales stage. Integration took weeks, the pilot expanded quickly across customer teams, and managers report more confident performance on real calls.
Delphi (digital minds): scaled live, real-time face-to-face video calls with Tavus CVI.
Final Round AI: scaled lifelike mock interviews for 100K+ users, logging 1.2M practice minutes with Tavus CVI.
CareFlick (AgeTech): combats senior loneliness with lifelike companions, chosen for realism and responsiveness.
Qualified: built Piper, its AI SDR, on Tavus, reporting 2X more conversations and 6X more meetings booked with qualified buyers.

Platform and methods

Speed and fidelity: sub-600ms utterance-to-utterance latency, real-time video, high-fidelity audio, and support for 43 languages.
Core models:
- Phoenix-4: real-time human rendering with built-in emotional intelligence. A full-duplex model that listens and responds at the same time.
- Raven-1: multimodal perception that fuses tone, prosody, expression, posture, and gaze into a single understanding of the person on the other side.
- Sparrow-1: transformer-based conversational timing that models the conversational floor continuously and resolves interruptions within tens of milliseconds.
Ethics and trust: SOC 2 Type 2, HIPAA, and GDPR compliance, with disclosure and consent mechanisms built in.
Pricing tiers: Free, Starter, Growth, and Enterprise, with included conversation minutes, faces, and scaling discounts.

Competitive advantages

Real-time realism: sub-600ms latency paired with rendering that carries genuine emotional expression.
Comprehensive stack: a full-stack system spanning perception, memory, knowledge, and action, built on Tavus’s own foundational models rather than assembled from third-party parts.
Flexibility: bring-your-own LLM, modular APIs, and a no-code builder for speed to market.
Scalability: from individual use cases to enterprise-grade deployments serving millions.
Ethical design: built with privacy, consent, and empathy at the core.

Ideal for

Product teams embedding face-to-face AI interactions into apps
Sales teams adopting AI or AI-assisted BDRs, SDRs, and sales callers
Consumers who need a powerful, functional, humanlike AI assistant that can accomplish tasks
Enterprises seeking scalable training, onboarding, and customer engagement
L&D and HR leaders replacing scripted training with interactive AI coaches
CX and Ops teams needing high-touch customer support without adding headcount
Individuals wanting on-demand AI tutors, mentors, or assistants

Content and thought leadership

The Tavus Blog: covers product launches, model research, responsible AI, and the future of human computing.

Trust signals

Venture-backed by CRV, Scale Venture Partners, Sequoia Capital, Y Combinator, HubSpot Ventures, and Flex Capital.
SOC 2 Type 2, HIPAA, and GDPR compliant.
Clear commitment to ethical AI: disclosure, privacy, and consent.
Trusted by 100,000+ developers and enterprises including Amazon, Deloitte, EY, Mayo Clinic, CVS, Salesforce, Aetna, and Wix.

See the glossary for commonly used terms, and how to use them.