Akool, a Palo Alto-based AI startup, has unveiled its latest innovation: Akool Live Camera, a tool that leverages artificial intelligence to revolutionize live video communication. The platform enables real-time translation, instant face swapping, and dynamic virtual avatars that mimic human movements and expressions—all with minimal latency.
Transforming Live Video with AI
Akool Live Camera captures human movement and instantly generates a photorealistic virtual avatar that mirrors the speaker’s gestures, tone, and facial expressions. The technology also translates speech in real time, synchronizing the translated audio with the avatar’s lip movements and emotional cues. This allows users to communicate seamlessly across languages during virtual meetings, live streams, and more.
“Our main motivation is to enhance the real-time experience,” said Jiajun “Jeff” Lu, CEO of Akool. “We want to make it so you can’t tell the avatar from the real person.”
Key Features
- AI Avatars: Lifelike avatars that reflect real-time expressions, gestures, and audience cues.
- Video Translation: Instantaneous, multilingual translation with synchronized lip-sync and voice identity preservation.
- Live Face Swap: Real-time face swapping with emotional accuracy, enabling users to represent different identities while maintaining authentic performance.
- AI Video Generation: Hyper-realistic, unscripted video generation on the fly, adapting to context and audience interaction.
Advanced Capabilities
- Unmatched Live Interaction: Real-time avatar streaming, face swap, and translation outperform traditional pre-recorded solutions.
- Dynamic Expression Mapping: Avatars mirror users’ emotions and movements for authentic engagement.
- Cross-Platform Integration: Compatible with Zoom, Microsoft Teams, Google Meet, and more.
- Privacy-First Design: Local facial data processing and professional avatars protect user identity in sensitive settings.
- Customizable Avatars: Choose from anime, retro, or business-centric avatars with flexible persona and outfit options.
Real-Time AI for Global Communication
Akool Live Camera’s edge-computing architecture processes live video feeds instantaneously, allowing avatars to adjust emotion, gestures, and speech cadence based on real-time audience analytics. This enables applications such as sales presentations in multiple languages, CEO addresses to global teams as digital avatars, and Twitch streams as animated characters—all with sub-100-millisecond latency.
“Akool Live Camera sets a new standard in AI-powered video generation technology, going well beyond scripted, text-based prompts,” said Lu. “This opens up a new array of possibilities for virtual meetings and live streams, especially when connecting with international audiences.”
Origins and Vision
Founded in 2022, Akool has rapidly expanded, serving clients like Coca-Cola and Qatar Airways. The company’s product suite includes video translation, real-time streaming avatars, studio-quality face swap, and talking avatars. Unlike text-prompt-based models such as OpenAI’s Sora, Akool’s technology thrives in unscripted, live environments, aiming to make synthetic humans indistinguishable from reality.
Akool’s team of around 80 people includes former Apple and Google engineers. CEO Jeff Lu brings experience from Google Cloud and Apple’s Face ID team. The company focuses on proprietary model development, from data collection to pipeline optimization, ensuring high performance and privacy compliance.
Ethics and Future Outlook
Akool prioritizes ethical AI development, embedding watermarks in AI-generated content and implementing content moderation tools. The company checks for copyright compliance during model training and processes sensitive data locally to enhance privacy.
Currently in beta and available to select early adopters, Akool Live Camera is slated for general availability in late 2025. The platform promises to redefine global communication, empowering creators, educators, and enterprises to connect more authentically and efficiently than ever before.
