San Mateo-based Kaon AI secured $60 million in funding to advance interactive entertainment using generative artificial intelligence, the company revealed Tuesday.
The investment round included participation from B Capital, Redpoint Ace, Goodwater Capital, and DCM. It follows an earlier $10 million pre-Series A round this year.
Previously known as FlowGPT, Kaon AI runs a platform that creates multimedia, character-driven stories in real time. Users guide the narrative through conversation, receiving customized visuals, dialogue, and video as the story progresses. Its main app, Emochi, reports over 2 million daily active users, with sessions lasting an average of 150 minutes.
Breaking away from static media
Jay Dang, Kaon’s CEO, stated that streaming platforms relying on pre-made content from fixed catalogs are outdated. He compared the current model to Blockbuster in the 2020s, calling it a relic. The company aims to build a new content engine that produces personalized narratives.
Traditional media depends on studios predicting audience tastes and monetizing those predictions. Kaon’s approach reverses this by generating content based on user input rather than matching users to existing material.
While other platforms like Character.ai have explored AI-driven storytelling, Kaon stands out by generating dynamic video and images alongside text. It also highlights “deep-preference memory,” which retains user choices across sessions, turning viewers into active participants in the storytelling process.
Related: Healthcare AI trust hinges on data foundations
Most interactive entertainment still resembles choose-your-own-adventure books, offering limited branching paths. Kaon argues that generative AI can eliminate those constraints, allowing users to steer stories freely while the system fills in visuals and dialogue. The challenge lies in maintaining coherence while supporting creativity.
Scaling the infrastructure
To support its goals, Kaon developed a custom technology stack running on over 1,000 GPUs through partners like Nebius Group and DigitalOcean. The setup processes trillion-level token volumes at about one-tenth the cost of traditional cloud services without sacrificing performance.
Kaon Labs, its research division, will analyze user data—including text inputs and narrative choices—to improve its AI models. The aim is to reduce the time needed to bring new interactive entertainment models to market from months to days.
Dang, a Berkeley graduate, leads a team of engineers and researchers specializing in real-time media generation. The $60 million will fund further platform development, though specific product timelines remain undisclosed.
Emochi remains the company’s main consumer product, offering a preview of AI-driven entertainment. Its success may hinge on whether users adopt stories that adapt in real time or grow weary of the concept. The balance between creativity and consistency will likely determine its long-term impact on media consumption.
As AI continues to reshape industries, validation platforms are also emerging to ensure security and reliability in evolving digital environments.
