Developers and companies are increasingly deploying AI agents and chatbots within their apps, but so far they’ve been mostly limited to text. Digital avatar production company Lemon slice is working to add a video layer to these conversations with a new diffusion model that can create digital avatars from a single image.
Called Lemon Slice-2, the model can create a digital avatar that runs on top of a knowledge base to play any role required of the AI agent, such as answering customer queries, helping with housework questions or even working as a mental health support representative.
“In the early days of GenAI, my co-founders started playing around with different video models and it became obvious to us that video was going to be interactive. The exciting part about tools like ChatGPT was that they were interactive, and we want video to have that level,” said co-founder Lina Colucci.
Lemon Slice says this is a 20 billion parameter model that can run on a single GPU for live video streaming at 20 frames per second. The company makes the model available through an API and an embeddable widget that companies can embed on their websites with a single line of code. After an avatar is created, you can change the background, style, and appearance of a character at any point.
In addition to human avatars, the company also focuses on being able to create non-human characters to suit different needs. The startup uses ElevenLabs’ technology to create the voices of these avatars.
Founded by Lina Colucci, Sidney Primas, and Andrew Weitz in 2024, Lemon Slice is betting that using its own general-purpose diffusion model (a type of generative model that learns to work inversely from noisy training data to generate new data) to generate avatars will set it apart from competitors.
“The existing avatar solutions I’ve seen to date add negative value to the product,” Colucci said. “They’re creepy and they’re stiff. They look good for a few seconds, and once you start interacting with them, it feels very weird and unsettling. What stopped the avatars from taking off was that they weren’t good enough.”
Techcrunch event
San Francisco
|
13-15 October 2026
To fund that effort, the company said Tuesday it raised $10.5 million in seed funding from Matrix Partners, Y Combinator, Dropbox CTO Arash Ferdowsi, Twitch CEO Emmett Shear and The Chainsmokers.
The company says it has safeguards in place to prevent unauthorized facial or voice cloning and uses large language models for content moderation.
Lemon Slice did not name the organizations using its technology, but said the model is being implemented for use cases such as education, language learning, e-commerce and corporate training.
The startup faces stiff competition from video production startups like D-ID, HeyGen, and Synthesia, as well as other digital avatar makers Genies, Soul Machine, Praktika, and AvatarOS.
Ilya Sukhar, partner at Matrix, believes that avatars will be useful in areas where videos are prominent. For example, people like to learn from YouTube instead of reading long sections of text. He noted that Lemon Slice’s technical prowess and his own will give it an edge over other startups.
“It’s a deep technical team with a history of shipping ML products, not just demos and research. Many of the other players are tailored to specific scenarios or industries, and Lemon Slice takes the general “Bitter lesson” scaling approach (of data and computation) that has worked in other AI methods,” he said.
Y Combinator’s Jared Friedman believes that using a diffusion-type model allows Lemon Slice to create any kind of avatar compared to some other startups that focus on either human avatars or game characters.
“Lemon Slice is, I believe, the only company that takes the fundamental ML approach that can finally overcome the uncanny valley and break the avatar Turing test. It trains the same type of model as Veo3 or Sora: a video diffusion transformer. Because it’s a general-purpose model that does everything end-to-end, it can’t see the upper limit to how well it works. It only needs one image to add a new face,” he said.
The startup currently has eight employees and plans to use the funds to hire engineers and go-to-market staff, along with paying the bills to train its models.
