Numz AI Raises 4 Billion Won for Structured Data Foundation Model Development

Led by Stonebridge Ventures
Participation from SBVA and KT Investment

Numz AI, a developer of structured data foundation models (TFM), announced on July 20 that it has raised its first investment of 4 billion won. The funding round was led by Stonebridge Ventures, with participation from SBVA, KT Investment, and Base Ventures.

Structured data refers to numerical data organized in table format with rows and columns. A significant portion of corporate data, such as sales records, transaction histories, and customer information, falls into this category.

The TFM developed by Numz AI is a foundation model for structured data. Just as a large language model (LLM) predicts the next word, TFM can infer empty values in a dataset with a single inference step. Traditionally, separate models for demand forecasting, anomaly detection, and credit evaluation have required several months for development; however, TFM can instantly process these tasks using a single model.

The structured data foundation model has recently garnered significant attention in the global artificial intelligence (AI) market. In February, U.S.-based Fundamental became the first AI unicorn focused on structured data, reaching a valuation of 1.4 billion dollars. In May, SAP announced the acquisition of Germany’s Prior Labs and a plan to invest over 1 billion euros over the next four years. Last month, Nvidia also acquired Kumo AI for approximately 400 million dollars.

Co-founders of Numz AI. From the left, Minyong Jo, CEO Jaemin Yoo, Dooho Lee, Co-founders. Photo by Numz AI

Numz AI was founded by Jaemin Yoo, currently an assistant professor in Seoul National University's Department of Computer Science and Engineering, along with fellow researchers. CEO Yoo was awarded the Google LhD Fellowship during his doctoral studies and was appointed as an assistant professor at KAIST at the age of 29. He has conducted research on AI for structured data—including graphs and time series—for over a decade.

Dooho Lee, co-founder, published three first-author papers at top AI conferences such as ICML and KDD within just 18 months of his master’s program. Minyong Cho, another co-founder, worked as a machine learning engineer at Dippinsource after completing his master’s degree at Seoul National University.

With this investment, Numz AI plans to expand its GPU infrastructure and research team to accelerate TFM development. The company also aims to broaden its business by providing on-premises solutions for enterprises with strict data transfer limitations, in addition to its API service, targeting industries such as finance, commerce, and healthcare.

Park Hyungsu, senior team leader at Stonebridge Ventures who led the investment, commented, "Most industrial decision-making happens not over language, but through tables and numbers. The transformations brought by LLMs in language are now beginning for structured data as well. In this emerging market where research capabilities set companies apart, we believe Numz AI has the potential to grow into a global leader."

Jaemin Yoo, CEO of Numz AI, stated, "Going forward, the ability to effectively utilize structured data will be a critical factor that determines a company's competitiveness. Leveraging our research capabilities and expertise, we aim to become a leading global AI company in the competition for foundation models."

This content was produced with the assistance of AI translation services.