Keyframe Labs: Turn agents into lifelike video calls. $0.06 a minute.

Turn agents into lifelike video calls. $0.06 a minute.

Our real-time, generative models turn AI into lifelike video calls. Deploy in minutes with a few lines of embed code. Photorealism, emotion, and scale for just $0.06 per minute.

Active Founders

Kaahan Radia

Founder

Kaahan is co-founder and CTO of Keyframe Labs, building the world's first photoreal, emotionally expressive AI humans for real-time video agents. Previously, he spent five years at Zipline as a Perception Engineer, building their camera-based detect-and-avoid system for autonomous drone operations. His work spans computer vision and applied ML, with research at CVPR and JAMA. B.A. CS, UC Berkeley.

Kaahan Radia

Founder

Kaahan is co-founder and CTO of Keyframe Labs, building the world's first photoreal, emotionally expressive AI humans for real-time video agents. Previously, he spent five years at Zipline as a Perception Engineer, building their camera-based detect-and-avoid system for autonomous drone operations. His work spans computer vision and applied ML, with research at CVPR and JAMA. B.A. CS, UC Berkeley.

Parth Radia

Founder

Parth is co-founder & CEO of Keyframe Labs. Before Keyframe, he spent 5 years at Ericsson on frontier AI, founding and leading a team of 30 building foundation models for voice & video synthesis on one of the world's first petascale human datasets, shipping Ericsson's first generative AI products. Before that, as a research scientist at AT&T Labs he developed novel neural architectures for photorealistic 3D telepresence, enabling the world's first 3D call over production networks. B.S. CS UCLA.

Parth Radia

Founder

Parth is co-founder & CEO of Keyframe Labs. Before Keyframe, he spent 5 years at Ericsson on frontier AI, founding and leading a team of 30 building foundation models for voice & video synthesis on one of the world's first petascale human datasets, shipping Ericsson's first generative AI products. Before that, as a research scientist at AT&T Labs he developed novel neural architectures for photorealistic 3D telepresence, enabling the world's first 3D call over production networks. B.S. CS UCLA.

Company Launches

Keyframe Labs - turn AI into lifelike video calls

See original launch post

tl;dr: our real-time, generative models turn AI into lifelike video calls. Deploy in minutes with a few lines of embed code. Photorealism, emotion, and scale for just $0.06 per minute.

We’re @Kaahan Radia and @Parth Radia from Keyframe Labs. Our interactive video models bring your AI agents and applications to life.

https://youtu.be/cIEP8OwTD4M

You can try a video call for yourself here: https://keyframelabs.com. To get a sense for the interactivity and expressive range, try asking for a sad Shakespearean monologue. Or for a “funny” dad joke.

But why video agents?

uploaded image

Face-to-face interaction triggers responses that text and voice simply can’t. Video calls close more deals, language learning with video tutors outperforms audio-only, and human presence drives outcomes in high-stakes interactions (e.g., telehealth, tax prep).

Now that voice has emerged as a primary conversational interface across industries, video is the next leap. Developers and enterprises already use Keyframe to easily deploy photoreal, conversational humans to a wide variety of use cases, including:

AI roleplay, for corporate training / learning and development
AI recruiting, for interview practice and automated screenings
Telehealth
High-touch customer support
AI sales and customer success reps
Always-on digital twins of influencers, to scale engagement and hyper-personalize interactions

How it works

uploaded image

Super simple.

Connect your agent or LLM, or choose from our supported providers
Select a persona from our stock library
Drop the embed code into your application

Need a custom persona or more control? Custom personas are available on enterprise plans, and our SDKs let you roll your own implementation.

Why Keyframe

Our research goals from day one have been:

Make meaningful progress towards beating the uncanny valley
Run at real-time, with world-scale efficiency

The result:

Industry-leading realism and concurrency limits
At a fraction of the cost of competitors

Get up and running

Sign up free (no credit card required) and build something cool
Have a use case? Book time with us directly
Know a company that could benefit from face-to-face AI interactions? We’d love an intro