Most organizations say they are not fully prepared to use AI genetic in a safe and responsible manner, According to recent McKinsey report. One concern is the explanation – the understanding of how and why AI makes certain decisions. While 40% of respondents consider it as a significant risk, only 17% are actively facing it, according to the report.
Based on Seoul Reference It started as a Data labeling company and now wants to help businesses create a safer AI with tools and data that allow the testing, monitoring and improving their models – without requiring technical expertise. On Monday, the start raised $ 15.5 million, which brings its total to about $ 28 million, from investors such as Salesforce Ventures, KB Investment, ACVC Partners and SBI Investment, among others.
David Kim, Managing Director of Datumo and a former AI researcher at the Korean Development Defense, was frustrated by the time -consuming nature of the data labeling to develop a new idea: an application based on rewards that allows anyone to point out their free time. The start ended the idea of a starting competition at Kaist (Korea Advanced Institute of Science and Technology). Kim founded Datumo, formerly known as Selectstar, along with five Kaist graduates in 2018.
Even before the application was fully constructed, Datumo secured tens of thousands of dollars in sales before the contract during the discovery phase of the competition customer, mainly from businesses led by Kaist graduate and newly established businesses.
In the first year, the start of $ 1 million revenue and secured several basic contracts. Today, the start counts large Korean companies such as Samsung, Samsung SDS, LG Electronics, LG CNS, Hyundai, Naver and Telecom Giant Sk Telecom based on its customers. A few years ago, however, customers began asking the company to overcome the simple data labeling. The 7 -year -old starts now has more than 300 customers in South Korea and has made about $ 6 million in revenue in 2024.
“They wanted to win their AI model outflows or compare them with other results,” said Michael Hwang, co -founder of Datumo, at TechCrunch. “Then we realized: We already did the evaluation of the AI model – without even knowing it.” Datumo doubled in this area and released Korea’s first database to focus on AI Trust and Safety, Hwang added.
“We started commenting on data and then expanded to preliminary data sets and evaluation, as the LLM ecosystem was matured,” Kim told TechCrunch.
TechCrunch event
Francisco
|
27-29 October 2025
META’s recent $ 14.3 billion investment in investments that resemble AI company AI underlines the importance of this market. Shortly thereafter, this deal, Ai Model Maker and Meta Competitor Openai, stopped using AI scale services. The META Agreement also marks that the competition for AI training data is intensifying.
Datumo shares some similarities with companies such as the AI scale in preparing the data set, and with Galileo and Arize AI in AI evaluation and monitoring. However, it differs through its licensed data, especially data from published books, which the company says it offers richly structured human reasoning, but it is known to be cleaned, according to CEO Kim.
Unlike its peers, Datumo also offers a full stack rating platform called Evaluation cardwhich automatically creates testing data and evaluations to check for unsafe, biased or incorrect answers without the need for manual beam of actions, Kim added. The signature product is a non -code for non -developers for non -developers such as politics, confidence and security and compliance groups.
When asked about attracting investors such as Salesforce Ventures, Kim explained that the start had previously hosted a conversation with Andrew Ng, founder of Deeplearning.ai, at an event in South Korea. After the event, Kim shared the LinkedIn session, which attracts the attention of Salesforce Ventures. After several meetings and zoom calls, investors expanded a soft commitment. The whole funding process took about eight months, Hwang said.
The new funding will be used to accelerate R&D efforts, in particular the development of automated evaluation tools for Enterprise AI, and for the escalation of GO-TO-MARKET global businesses across South Korea, Japan and the US, which has 150 employees in Seoul. March.
We always try to evolve and by providing some image of your perspective and feedback on TechCrunch and our coverage and events, you can help us! Complement this investigation To let us know how we do and have the opportunity to win a prize in return!
