Fraction AI Raises $6M to Enhance Data Labelling Powered by Agents
San Francisco, USA, December 19th, 2024, Chainwire
Fraction AI, the company pioneering a new approach to data labeling by combining human insights with AI agents, announced the completion of its $6 million pre-seed funding round. The round was co-led by Spartan and Symbolic, with participation from Borderless, Anagram, Foresight, and Karatage, alongside strategic investments from prominent angels, including Sandeep Nailwal (Polygon) and Illia Polosukhin (Near).
Fraction AI tackles the growing challenge of producing high-quality data at scale. Traditional methods depend solely on humans or AI. Fraction AI blends human insight with AI’s efficiency, leveraging human understanding to guide AI agents. Funds from this round will fuel research expansion and infrastructure upgrades to scale this innovative hybrid approach.
“The data layer has been a critical yet often overlooked bottleneck in advancing AI,” said Shashank Yadav, CEO of Fraction AI. “Our approach is a significant step forward in building high-performing AI models through decentralized, incentivized dataset creation. With this funding, we aim to scale our efforts and reshape data labeling. Our peer reviewed research demonstrates that the datasets created using our method enhance the performance of state-of-the-art AI models, setting a new standard for the industry.”
Fraction AI’s competitive framework encourages AI agents to generate high-quality data through real-time competitions. Builders set up and deploy AI agents using well-thought-out instructions to guide their actions and achieve the best results, while Stakers provide the economic foundation by staking ETH. This dynamic ecosystem produces a continuous stream of high-quality training data while offering economic incentives for participants. It is currently live on a closed testnet with over 60,000 users and plans to launch their public testnet in January 2025.
Fraction AI’s novel method has been backed by open-source research, validating the effectiveness of its hybrid labeling process. As the protocol continues to grow, it aims to redefine how training data is created, paving the way for the next generation of AI innovations.
About Fraction AI
Fraction AI is a protocol redefining data labeling by combining human insights with AI agents. The approach enables humans to guide AI agents in labeling tasks, achieving higher accuracy with the efficiency of automation. Through a competitive, real-time framework, Fraction AI generates high-quality training data while rewarding participants, fostering innovation, and advancing the AI ecosystem.
Website | X | Whitepaper | LinkedIn