Artificial intelligence startup Lemon Slice on Tuesday launched Lemon Slice-2, a real-time AI avatar model designed to turn a single image into a live, conversational video call, as companies race to make digital interactions more immersive.
The company said the system works in a zero-shot format, allowing users to upload one image and immediately begin a real-time video interaction, without the need for training data or pre-recorded footage. Lemon Slice-2 is being released both as an application programming interface for developers and as an embeddable widget that businesses can add to their websites.
See also: Google Translate Adds Live Speech Translation Via Headphones
According to Lemon Slice, the model is built on a proprietary large-scale video diffusion transformer, similar in class to video generation systems developed by larger technology firms but optimized for real-time conversational use. The technology generates all visual elements from scratch, enabling expressive facial animation, gestures and full-body movement during live interactions.
The startup said the product is intended for a wide range of use cases, including e-commerce assistance, online education, healthcare intake and coaching, where interactive visual agents may improve engagement compared with text-based chatbots.
See also: OpenAI Expands ChatGPT Image Capabilities With New Model
Lemon Slice said it has implemented safeguards to prevent misuse, including consent requirements, content moderation and clear disclosure that users are interacting with an AI character. The company has raised $10.5 million in seed funding from investors including Matrix Partners and Y Combinator to support further development and commercialization.
Source: PR Newswire