Banseog note · 2026-10-26
A company building robot brains from data on how people work
Current stage · from public events · 2026-11-14
- R&D · prototype
- Pilot · first production
- Scaling production
- Operations · field rollout
- Going global · localizing
- M&A · integration
- Restructuring
Pilot · first production Validating action data and robot models with customers and robot manufacturers
- 2026-10-14 In an interview, the company stated that, alongside developing robot data and models for customers, it had conducted technical validation this year with U.S.-based Dexmate. Source
Roles often seen at this stage Robot software, AI & vision · Controls & embedded · Field, integration & service
Companies at the same stage → · Same stage in other industries →
- A robot data and AI company founded with bases in Silicon Valley and Seoul, it raised USD 27 million (about KRW 40 billion) in a seed round led by Samsung Venture Investment in May 2026, 16 months after its founding. Total funding including the pre-seed is about USD 35 million (about KRW 51.5 billion).
- It said that as of October 2026 it had collected more than 150,000 hours of behavior data (more than 400,000 tasks and more than 10 million episodes), and it gathers about 20,000 hours a month at its data factory in Hanoi, Vietnam.
- It said that by adopting AWS infrastructure it cut robot training costs in half and data transfer costs to one-sixth, and that it processes more than 100 TB of data a month.
Decisions within a year
- USD 27 million seed round (May 2026): led by Samsung Venture Investment, with participation from Mirae Asset Venture Investment, Korea Development Bank, LG Technology Ventures, SKT Capital Ventures, Hyundai ZER01NE Ventures, Kakao Ventures and others. The company said it plans to increase GPU investment, hiring and R&D.
- Development of the robot foundation model CFG-1 (unveiled October 2026): trained on more than 100,000 hours of its own human work data, it highlights a memory function that lets the robot remember tasks it has done before.
- Industry partnerships it said it is working with LG CNS and Samsung SDS on robot transformation (RX) projects and is running a proof of concept (POC) in 2026 with US robot company Dexmate. It also said it has Korean customers in semiconductors, automotive and systems integration.
People they will look for
- Robot foundation model researcher — training models on large-scale behavior data in step with advancing CFG-1 and expanding GPU investment
- Robot data platform and infrastructure engineer — connecting the more than 100 TB a month coming in from the Hanoi data factory into cleaning, storage and training pipelines
- Robot application solutions engineer — applying models to real robots in the LG CNS and Samsung SDS partnership projects and at semiconductor and automotive customer sites
For hiring managers
People who can handle robot foundation models and large-scale data infrastructure are also sought by the AI research units of large Korean companies, the Korean labs of big tech firms and other physical AI startups. Now that the company has announced expanded hiring after a large seed round, securing key research talent first is the advantage.
For job seekers
Experience building robot learning (imitation learning, VLA models) or large-scale training data pipelines yourself, along with experience putting models onto real robots, is what this company values most right now.
An analysis written from public sources. Sources: The Korea Economic Daily 2026-05-11 (USD 27 million seed, total funding, founding bases), Seoul Economic Daily (English edition) 2026-10-14 (150,000 hours of data, CFG-1, LG CNS, Samsung SDS, Dexmate, hiring plans), ZDNet Korea 2026-05-21 (AWS cost savings, 100 TB a month).
Open jobs
Banseog checks official company careers pages every day. A job is marked closed only when the official page returns a removal (404 or 410).