Datumo, a South Korean AI startup, has raised $15.5 million in a fresh funding round to scale its large language model (LLM) evaluation platform aimed at bolstering AI safety and trust. Led by Salesforce Ventures, the round included participation from KB Investment, ACVC Partners, SBI Investment, and others, bringing Datumo’s total funding to around $28 million.
From Data Labeling to Trust-Focused AI Tools
Founded in 2018 by AI researcher David Kim and five KAIST alumni, Datumo initially built a reward-based data labeling app that caught early attention. The company earned over $1 million in revenue in its first year and quickly onboarded high-profile clients including Samsung, LG, Hyundai, Naver, and SK Telecom.
As clients increasingly asked Datumo not just for labels but help testing and comparing AI model outputs, it pivoted. The startup launched Datumo Eval, a no-code evaluation platform enabling policy, compliance, and trust teams to test AI models for bias, accuracy, and safety—without deep technical expertise.
Addressing a Growing Demand for Explainable AI
Datumo’s push comes at a critical juncture: a recent report revealed only 17% of organizations are actively addressing AI explainability, despite its wide recognition as a key concern. Datumo’s platform bridges this gap by providing automated test data generation and comprehensive evaluation dashboards, serving over 300 enterprise clients in Korea and generating approximately $6 million in revenue in 2024.
The timing of the funding mirrors broader market movements—such as Meta’s multi-billion-dollar investment in Scale AI, underscoring high demand for reliable data evaluation services. Datumo differentiates itself with licensed datasets derived from published sources, rich in structured human reasoning but historically difficult to work with.
Strategic Scaling and Global Expansion
The fresh capital will fuel Datumo’s research and development in enterprise AI evaluation tools, particularly its Datumo Eval platform. Plans are also underway for expansion in Japan and the U.S., leveraging its established Silicon Valley presence since March 2025. With a team of roughly 150 based in Seoul, Datumo aims to emerge as a trusted global provider of AI safety infrastructure.
Why Datumo Matters
- Accessibility for Non-Technical Teams: Datumo Eval democratizes AI testing, allowing non-developers to validate model behavior.
- Sovereign AI Ecosystem: Its proprietary datasets and local focus challenge Western-dominated AI infrastructure.
- Response to Industry Pressure: As regulatory scrutiny of AI grows, Datumo offers organizations tools to ensure transparency and trust.
Remaining Challenges
Despite its promising trajectory, Datumo must address pressing challenges:
- Scaling Accuracy: Ensuring its automated evaluation maintains reliability across diverse model types and domains.
- Global Market Familiarity: Building trust among international users accustomed to entrenched AI evaluation platforms.
- Competitive Landscape: Staying ahead of larger players expanding into trust and safety tooling.
Looking Forward
With generative AI becoming integral to enterprise operations—from finance to healthcare—the need for safe, explainable systems has never been clearer. Datumo’s no-code evaluation tools and unique data approach position it well for broader adoption. As the platform matures and expands globally, it may well become a critical ally for businesses deploying AI responsibly.