OriginFlow
OriginFlow builds embodied-AI data infrastructure based on human manipulation signals. The company uses wearable sensing and human-motion capture to turn real human skill into training data for robots, supporting physical-AI systems that need scalable demonstrations for dexterous work.
Market context
OriginFlow is relevant because dexterous robot behavior is constrained by data quality, not only hardware. The company is building a data layer for fine manipulation, home-service skills and physical interaction datasets that robot builders can use to train more capable systems.
Facts
- Website: https://www.originflow.ai/
- HQ: Beijing, China
- Founded: 2025
- Segment: physical-AI data infrastructure
- Known funding rounds: 3
Investors
- Guofang Venture Capital - Participant - Vc - 1 funding event - first seen 2026-05-21
- Monolith Capital - Lead - Vc - 1 funding event - first seen 2026-05-21
- Puhua Capital - Participant - Vc - 1 funding event - first seen 2026-05-21
- Yuanhe Origin - Participant - Angel / Individual - 1 funding event - first seen 2026-05-21
- Yuanhe Puhua - Participant - Angel / Individual - 1 funding event - first seen 2026-05-21
Embodied-AI data layer for dexterous robot learning
OriginFlow builds data infrastructure for embodied AI, using human manipulation and neuromuscular signals to produce higher-quality robot skill data.
- Target environment: Robot learning labs, humanoid and mobile-manipulation teams, home-service robotics developers, human-machine interaction workflows and embodied-AI data pipelines.
- Deployment model: Infrastructure provider model built around embodied datasets, neuromuscular sensing, model encoding and data services for robot training rather than finished robot hardware.
- Customer context: Robot developers, embodied-AI labs, home-service robot companies, manipulation researchers and teams that need hard-to-collect dexterous interaction data.
- Workflow context: Human manipulation capture, sEMG signal acquisition, hand pose and force reconstruction, haptic feedback estimation, robot skill dataset generation and home-service task data collection.
- Commercial maturity: Early-stage company with multiple 2026 financing tranches and public positioning around NeuroScale, PULSE and embodied-AI data infrastructure.
- Market position: Robot data infrastructure company focused on the scarce-data layer beneath dexterous manipulation and home-service robotics.
- Adoption constraints: Adoption depends on data quality, sensor reliability, task coverage, privacy and consent controls, robot-platform transfer, model performance and proof that human signal data improves deployed robo…
- Adjacent context: Embodied-AI datasets, robot foundation models, manipulation learning, teleoperation data, dexterous hands, home-service robots, haptic sensing and human-machine interaction.
- Source confidence: medium
Events
- 2026-05-21 - Funding - OriginFlow closes Pre-A1 round led by Monolith Capital - The Pre-A1 round was led by Monolith Capital. Puhua Capital made a follow-on investment, with Yuanhe Puhua, Yuanhe Origin and Guofang Venture Capital also named in coverage. The capital supports commercial automation and autonomous-system workflows as OriginF…
- 2026-03-01 - Funding - OriginFlow adds strategic and industrial investors in strategic financing round - The strategic round brought in 58 Strategic Investment, Puhua Capital and the Shuimu Tsinghua Seed Alumni Fund. Existing shareholders including BlueRun Ventures and Vitalbridge made follow-on investments described as tens of millions of yuan. The exact close…
- 2026-01-01 - Funding - OriginFlow closes angel round co-led by BlueRun Ventures and Vitalbridge - The angel round was co-led by BlueRun Ventures and Vitalbridge. The exact close date and standalone amount were not disclosed; this timeline date marks the first named tranche in the reported sequence. The financing supports OriginFlow's NeuroScale approach f…
Team
- Qin Shentao - Founder and CEO - Leads OriginFlow's embodied-AI data and NeuroScale strategy for turning human manipulation signals into robot skill data.
Context tags
- Tactile robotics data - Stack Layer - OriginFlow develops NeuroScale data capture for embodied-AI physical interaction data.
OriginFlow canonical Korthos profile