Story-based AI data layer 'Poseidon' launches training data collection app

STORY, a blockchain infrastructure for intellectual property (IP), announced on the 3rd that it has officially launched a training data collection app for the physical AI operations of its incubating project, the decentralized AI data infrastructure 'Poseidon'.

This is Poseidon's first app and presents a new data economy model in which anyone worldwide can participate in AI training data collection and receive rewards.

In July, Poseidon attracted industry attention by securing $15 million (about 20.8 billion won) in seed funding led by the global venture capital firm Andreessen Horowitz (a16z Crypto).

Poseidon collects high-quality datasets that are difficult to obtain through traditional internet crawling methods, such as first-person view (POV) videos for robot training and multilingual voice, and has built infrastructure for labeling and licensing them. The collected datasets go through preprocessing, are registered as IP on the Story blockchain, are transparently tracked and distributed, and royalties are automatically allocated. AI companies can use them by paying appropriate fees and utilize the data without copyright issues.

Poseidon will secure sensor data by linking with apps developed by multiple decentralized physical infrastructure network (DePIN) partners, and plans to sequentially unveil an infrastructure stack including a software development kit (SDK) and integrated licensing tools. The app launched this time is designed to make it easy for users to participate in data collection, aiming to secure more than 1,000 voice datasets per non-English language such as Korean, Hindi, and Japanese.

Users can earn points by completing quests in which they read scripts provided in the app in the target language and record their voice. The accumulated voice data helps AI models understand human language more diversely and accurately.

Starting from this, Poseidon plans to add apps involving external developers and collect various high-quality data such as text and video.

The Poseidon team has deep expertise in AI infrastructure centered on Sandeep Chinchali, STORY's Chief AI Officer (CAIO) and the project's lead scientist. Sandeep earned a Ph.D. in AI and distributed systems from Stanford University and is currently a professor at the University of Texas at Austin (UT Austin), researching AI, robotics, and distributed systems. Sarick Shah, head of product and co-founder, is a machine learning engineer with over 10 years of experience in scaling AI products for large enterprises in finance, telecommunications, and healthcare.

Seung-yoon Lee, CEO of Story and chair of Poseidon, said, "AI companies can no longer rely only on public datasets, and the need for new forms of data infrastructure is growing. Poseidon collects data in a decentralized manner through a blockchain network that anyone can participate in, and contributors receive fair distribution of rewards. Through this, AI companies can utilize high-quality datasets that are difficult to obtain from the internet without copyright issues."