Korean AI startup Numz AI raises $ KRW 2bn, totaling 6bn won
Numz AI, a Seoul-based developer of tabular data foundation models, has closed a 2 billion won second pre-seed round, lifting its total early funding to 6 billion won.
SUMO· the number puzzle
Dialo· learn languages by speaking
ediviewer· read EDI files like documents
giftdetect· the right gift in minutesNumz AI, a South Korean startup building foundation models for tabular data, has raised 2 billion won (about 10 billion won from each of two investors is not the case: 1 billion won each) from Mirae Asset Venture Investment and Korea Investment Partners in a second pre-seed closing. With the July first round included, the company has secured 6 billion won in nine months since its incorporation.
Investors and public funding
The first closing brought in 4 billion won from Stonebridge Ventures, SBVA (formerly SoftBank Ventures), KT Investment and Base Ventures. Numz AI has also been selected for the Ministry of SMEs and Startups' 2026 third-round TIPS R&D general track, operated by Base Ventures, which adds 800 million won in research funding. The company plans to spend the money on improving its tabular-data AI model and hiring core engineers.
What Causilo does
The company's proprietary model, Causilo, analyzes row-and-column numeric data such as sales ledgers, payment logs and customer behavior histories to produce business forecasts, including churn prevention and demand prediction. In September 2026 it ranked first overall, and in the classification and regression categories, on the TabArena benchmark for tabular data AI. In October, Nvidia CEO Jensen Huang presented it during the GTC Berlin 2026 keynote, alongside Nvidia's Kumo-Tabular and Germany-based Prior Labs' TabPFN-3.5, as one of the key global models in the field.
Numa AI was founded in February by a team of researchers from Seoul National University and KAIST, led by Seoul National University computer science professor Yoo Jae-min, as a faculty startup. It is currently working with major Korean manufacturing and platform companies, including InBody, on decision-making projects. The firm has open-sourced Causilo's base code and pretrained weights, while offering customized commercial licenses to enterprise customers.
What comes next
CEO Yoo said competition among big tech firms in tabular foundation models is heating up, and that the speed of technical refinement and market capture will decide the winners. He said the new capital and TIPS support will go toward improving Causilo's computational efficiency and predictive performance to accelerate its push into global markets.
Apps worth a look
SUMO: the number puzzleTap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.
Dialo: learn languages by speakingTalk to an AI conversation partner; every sentence comes back corrected and explained in your language. 9 languages, A1 to C2.
ediviewer: read EDI files like documentsOpen EDIFACT and X12 messages, see every segment explained, convert to JSON/XML, export PDF and Excel. Runs locally, no upload.
giftdetect: the right gift in minutesPick who it is for, the occasion and their personality. The search engine reads product descriptions from local shops and ranks what fits. Free, no sign-up.
More from this day
See all →
Pennylane, Indy and Tiime lead France's e-invoicing connections
South Korea to Fully Lift Network Separation Rules for Banks
Elsevier Adds LG AI Research Vision Model to Reaxys Search
