The Seoul -based Datumo raises $ 15.5 million to take AI, supported by Salesforce

Most organizations say they are not fully prepared to use gynecological artificial intelligence in a safe and responsible way, According to McKinsey’s latest report. Anxiety is clarification – understanding how and why artificial intelligence takes certain decisions. While 40 % of the respondents view it as a great danger, only 17 % treat it actively, according to the report.

It is based in Seoul Datumo It has started as an AI data mark laying company, and you now want to help companies build more secure artificial intelligence with tools and data that allow testing, monitoring and improving their models – without asking for technical experience. On Monday, the startup raised $ 15.5 million, raising its total to about 28 million dollars, including investors including Salesforce Ventures, KB Investment and Sbi Investment, among others.

David Kim, CEO of Datumo and a former Korea Defense Development Agency, was disappointed by the time consuming time to put signs of data, so he reached a new idea: a reward -based application that allows anyone to nominate data in his spare time and earn money. The startup company has verified the idea in the startup competition in Kaist (Korea Institute Advanced Science and Technology). Kim participated in the founding of Datumo, previously known as Selecstar, along with five Kaist graduates in 2018.

Even before the construction of the application completely, Datumo got tens of thousands of dollars in pre -contract sales during the client’s discovery phase in the competition, most of them are companies led by graduates and startups.

In its first year, the startup company exceeded a $ 1 million revenue and received many major contracts. Today, the startup company is the major Korean companies such as Samsung, Samsung SDs, LG Electronics, LG CNS, Hyundai, Naver and SK SK, which is based in Seoul, which is based in Seoul among its customers. Several years ago, customers started asking the company to bypass a sign of simple data. The seven -year -old emerging company includes more than 300 customers in South Korea and achieved revenues of about $ 6 million in 2024.

“They wanted us to record their artificial intelligence outputs or compare them to other outputs,” Michael Huang, co -founder of Datumo, told Techcrunch. “This is when we realized: We were already evaluating the artificial intelligence model – without knowing it.” Huanga added that Datumo has doubled in this field and issued the first supportive group of data for Korea, focusing on the confidence of artificial intelligence and safety.

“We started explaining the data, then we expanded to pre -training and evaluation groups with the maturity of the LLM ecosystem,” Kim told Techcrunch:

Investment similar to the acquisition of $ 14.3 billion in Meta in the company’s domain of AI It highlights the importance of this market. Shortly after that, the maker of artificial intelligence and descriptive competition Openai stops using scale ai services. The Meta deal also indicates that competition for artificial intelligence training data is condensed.

Datumo shares some similarities with companies like Scale AI in providing pre -training data set, and with Galileo and Arize ai In the evaluation of artificial intelligence and monitoring. However, it distinguishes itself through licensed data collections, especially the data that crawled from the published books, which the company says provides a rich human logic but is difficult to clean, according to CEO Kim.

Unlike its peers, Datumo also offers a full -fledged evaluation platform called Datumo EvalKim added that he is the one who automatically creates test data and evaluations to verify unsafe, biased or incorrect responses without the need for hand programming. The distinctive product is an evaluation tool without a symbol designed for non -developers such as those in the teams of politics, trust, safety and compliance.

When asked about attracting investors such as Salesforce Ventures, Kim explained that the startup has previously hosted Fireside chat with Andrew NG, founder of Deeplearning.ai, at an event in South Korea. After the event, Kim shared the LinkedIn session, which caught the attention of the Slesforce projects. After many meetings and magnification calls, investors have expanded a delicious commitment. Huang said that the entire financing process took about eight months.

The new financing will be used to accelerate research and development efforts, especially in developing Enterprise AI, and expanding the scope of going to the global market across South Korea and Japan, and the United States, which includes 150 employees in Seoul, also established a presence in Silicon Valley in March.