Datumo Raises $15.5M to Expand Safe AI Model Evaluation Globally
Datumo Secures $15.5M Funding to Advance AI Trust and Safety Tools
South Korean AI startup Datumo has announced a $15.5 million investment round, boosting its total funding to approximately $28 million. The round saw participation from major investors including Salesforce Ventures, KB Investment, and SBI Investment. This marks a significant step for Datumo as it aims to help enterprises safely deploy generative AI models.
Addressing AI Safety and Explainability
According to a recent McKinsey report, most organizations are not fully prepared to use generative AI responsibly, with explainability being a core concern. While 40% of surveyed leaders consider AI explainability a major risk, only 17% are actively working to address it. Datumo’s mission is to bridge this gap through user-friendly tools for AI testing, monitoring, and improvement—no deep tech expertise required.
From Data Labeling to Full AI Evaluation
Founded in 2018 by CEO David Kim (a former AI researcher at Korea’s Agency for Defence Development) and five KAIST alumni, Datumo began as an AI data labeling company. Kim’s idea of a reward-based data labeling app quickly gained traction, generating pre-contract sales even before the product launched. In its first year, Datumo topped $1 million in revenue, signing contracts with leading Korean tech giants such as Samsung, LG, Hyundai, Naver, and SK Telecom.
As client needs evolved, Datumo expanded beyond data annotation. Co-founder Michael Hwang noted, “They wanted us to score their AI model outputs or compare them to other outputs. That’s when we realized: we were already doing AI model evaluation — without even knowing it.” Building on this, Datumo released Korea’s first benchmark dataset focused on AI trust and safety, supporting the responsible adoption of large language models (LLMs).
Innovation in AI Model Testing
Datumo now serves over 300 clients in South Korea and reported $6 million in revenue for 2024. The company stands out for its extensive licensed datasets, including data crawled from published books—valuable for structured human reasoning but difficult to clean. Datumo also offers a full-stack evaluation platform, Datumo Eval, which automates test data generation and evaluation. This no-code tool empowers non-developers in policy, trust and safety, and compliance roles to check for unsafe, biased, or incorrect AI responses easily.
Competing in a Growing Market
The importance of reliable data and AI evaluation has grown rapidly, emphasized by Meta’s recent $14.3 billion investment in Scale AI and subsequent industry shifts. Datumo competes with established players like Scale AI, Galileo, and Arize AI, but differentiates itself through proprietary datasets and a focus on accessible, automated evaluation tools.
Global Expansion and Future Plans
Investor interest was partly sparked by Datumo’s collaboration with prominent figures like Andrew Ng and a series of high-profile events. According to Kim, the company will use its new funding to accelerate R&D, enhance automated AI evaluation tools for enterprises, and expand its market reach across South Korea, Japan, and the U.S. Datumo, with 150 employees and a new presence in Silicon Valley, is poised to become a key player in making AI safer and more transparent worldwide.
