DEEP RESEARCH · SELECTSTAR/DATUMO
SelectStar Deep Investment Analysis Report
A data and evaluation infrastructure company for the age of AI reliability regulation
0. Bottom line first
SelectStar’s core shift is from data labeling to AI reliability evaluation. I see the case in Korea’s AI Basic Act, the EU AI Act, financial and public-sector compliance, and the SaaS transition around Datumo Eval.
Official fact: The source describes SelectStar under the global brand Datumo and highlights Korea’s AI Basic Act in 2026, the EU AI Act, the 2026-2028 growth outlook, and a 2026 technology-special listing goal.
Interpretation: As AI competition moves from raw performance toward reliability, safety, and auditability, SelectStar’s third-party evaluation position becomes more valuable.
1. From Data 1.0 to AI Reliability 2.0
SelectStar was founded in November 2018 by a KAIST-origin team including CEO Se-yup Kim and co-founder Ho-wook Shin. The initial problem was the data bottleneck in AI development.
Cashmission built a mobile crowdsourcing pool with more than 320,000 cumulative users and supported AI training-data projects for companies such as LG CNS, Samsung Electronics, and SK Telecom. The source emphasizes cross-validation and proprietary quality algorithms that maintain data quality above 99%.
2. Investment points
Regulatory moat
Korea’s AI Basic Act, expected in January 2026, requires reliability measures for high-impact AI.
Salesforce Ventures
Participation in Series B is presented as a signal for North American and Japanese B2B expansion.
Datumo Eval
The business is shifting from labor-intensive projects to recurring, higher-margin software revenue.
3. Funding and business model
| Item | Source detail | Meaning |
|---|---|---|
| Cumulative funding | About KRW 43.4B as of Dec. 2025 | Strong capital base in AI data/evaluation |
| Series A | About KRW 4B in 2020-2021 | Cashmission and sales expansion |
| Series B | KRW 20.5B in Aug. 2025 | Salesforce Ventures, KB, Shinhan, SBI and others |
| Dataset licensing | Started in 2023; cumulative sales above KRW 5B | Move from projects to asset-based revenue |
The moat is a combination of humans and software: domain experts in legal, medical, and finance plus automated evaluation and vulnerability testing through Datumo Eval.
4. Competition and roadmap
The source compares SelectStar with Scale AI, Crowdworks, and AIMMO. Scale AI has global big-tech relationships, while SelectStar differentiates with Korean-language nuance, local regulation, and on-premise/private-cloud deployment.
- 2026: Win financial and public compliance projects and pursue a KOSDAQ technology-special listing.
- 2027: Launch Datumo Eval on Salesforce AppExchange and target North American enterprise customers, with SaaS subscriptions above 40% of revenue.
- 2028: Move into real-time AI-agent safety monitoring and control as an Asia No.1 AI Safety Platform candidate.
5. Risks and final view
Risks
- Burn rate from platform development and global expansion must be managed.
- Big tech platforms such as Naver and Kakao may internalize AI evaluation tools.
- If AI Basic Act enforcement is softened, expected revenue acceleration may be delayed.
My conclusion is that SelectStar has evolved into a precision instrument for checking AI engines. It is an attractive unlisted AI infrastructure candidate around a 2026 IPO, but the actual regulatory bite and SaaS revenue conversion must be verified.