Software development
Worldmodeldata lands £7M to turn gaming data into AI training
Worldmodeldata emerges from stealth with funding to expand its platform for licensed video game training data for AI applications in robotics, autonomous vehicles and beyond.
Tamara Djurickovic12 hours ago
Cambridge-based
startup Worldmodeldata has raised £7 million in seed funding as it emerged from
stealth. The round was led by Iona Star Capital, a London-based venture capital
firm focused on early-stage companies in artificial intelligence, data and
technology
Founded
by serial entrepreneur Rhea Loucas, the company is building a database of video
game-generated training data for AI systems known as world models, which are
designed to understand and predict how environments change over time. Lord
Richard Allan, former vice president of public policy at Meta and a UK
technology policy specialist, has joined the company’s board as chairman
Worldmodeldata
aggregates and structures gameplay data from modern video games to create
datasets for organisations developing world models, physical AI systems and
robotics. The data isunities, including titles built on Unreal and Unity, rather than
through web scraping
The
platform targets what the company describes as a growing shortage of
high-quality training data for AI systems that need to understand complex
environments and make decisions in real-world settings. Potential applications
include autonomous vehicles, where world models can be used to simulate traffic
conditions and predict pedestrian movement
Rhea
Loucas said the company was founded to
address the growing need for large-scale training data for world models:
World
models represent a significant shift in AI, but they require large-scale
datasets that enable systems to understand and reason about physical
environments. Video games provide rich, controlled environments that can
generate the data needed to train these models, and our goal is to make that
data available at scale
The
funding will support product development, team expansion and new data licensing
agreements as Worldmodeldata works towards building a library of one million
hours of training data by the end of next year
Would you like to write the first comment?
