The archive · AI & Models · Strategic decision · 2018–2026
Datumo bets AI evaluation is a no-code enterprise category; $15.5M Salesforce round
Seoul's Datumo pivoted from data labeling to no-code LLM evaluation, raised $15.5M (total ~$28M) and serves Samsung, LG, Hyundai, Naver and SK Telecom
Datumo (formerly SelectStar)
What the business is
A Seoul-based AI data and evaluation company: crowdsourced data labeling, licensed pretraining datasets, and Datumo Eval, a no-code platform that scores and monitors LLM outputs for unsafe, biased or incorrect responses.
Starting capital：~$28M total raised through August 2025, including the $15.5M Series B
How it started
David Kim, a former AI researcher at Korea's Agency for Defense Development, found data labeling painfully slow and founded Datumo (then SelectStar) in 2018 with five KAIST alumni. The crowdsourced labeling app passed $1M revenue in its first year and built a base of enterprise clients in Korea.
What happened
As the LLM ecosystem matured, clients began asking Datumo to score their model outputs and compare them to alternatives. The company doubled down: it launched Datumo Eval, a full-stack no-code evaluation platform that auto-generates test data and checks for unsafe, biased or incorrect responses, and released Korea's first benchmark dataset focused on AI trust and safety. The Aug 2025 $15.5M Series B, which brought total funding to about $28M, was earmarked for R&D in automated evaluation tools and global go-to-market in South Korea, Japan and the US.
How it ended up
Still scaling: as of August 2025 Datumo had 150 employees in Seoul, a Silicon Valley office opened in March 2025, and it was positioning Datumo Eval against evaluation/monitoring players like Galileo and Arize AI rather than competing only on data.
Background
Datumo is a Seoul-based AI data and evaluation company that began as a data-labeling startup and pivoted into no-code LLM evaluation as the generative-AI market matured. Its signature product, Datumo Eval, automatically generates test data and checks AI model outputs for unsafe, biased or incorrect responses — designed for non-developers on policy, trust-and-safety and compliance teams.
Founded in 2018 as SelectStar by David Kim and five KAIST alumni, Datumo validated its crowdsourced labeling app at a KAIST startup competition, passed $1 million revenue in its first year, and grew to 300+ clients in South Korea including Samsung, LG, Hyundai, Naver and SK Telecom, generating about $6 million revenue in 2024.
In August 2025 Datumo raised a $15.5 million Series B — bringing total funding to about $28 million — with Salesforce Ventures, KB Investment, ACVC Partners and SBI Investment participating. The capital was aimed at automated enterprise evaluation tools and expansion across South Korea, Japan and the US.
The company frames itself as a challenger to Scale AI in pretraining data and to Galileo/Arize AI in evaluation, differentiated by licensed book-sourced datasets and a no-code evaluation platform that lets non-engineers score and monitor AI systems.
What has to be true
- The LLM boom made model evaluation and trust/safety an urgent enterprise need — McKinsey data shows most organizations aren't prepared for safe generative AI use.
- Buyers sit on policy, compliance and trust teams, not engineering teams, so no-code evaluation tools serve a bigger addressable audience than developer frameworks.
- A decade of data-labeling client relationships gave Datumo distribution and a natural discovery channel for the pivot.
- Meta's ~$14.3B investment in Scale AI signaled that AI data and evaluation is a strategic, well-funded category.
What can be applied
A services customer base is a discovery channel: Datumo didn't invent an evaluation product in a vacuum — annotation clients asked for it, which made the pivot lower-risk and instantly sold.
Aftermath
As of August 2025 Datumo was scaling: it had 150 employees in Seoul, had opened a Silicon Valley office in March 2025, and was expanding go-to-market across South Korea, Japan and the US. Its ~$28M total funding and 300+ Korean enterprise clients positioned it as one of the leading domestic players in LLM evaluation as that category consolidated.
Sources
- Seoul-based Datumo raises $15.5M to take on Scale AI, backed by Salesforce
- Seoul-Based Datumo Raises $15.5 Million to Challenge Scale AI, Backed by Salesforce
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card