BlueRun Ventures Leads OriginFlow's Angel Round, OriginFlow Tackles Embodied Intelligence Data Bottleneck with NeuroScale Paradigm | BlueRun Family

Recently, embodied intelligence data technology company OriginFlow has completed multiple consecutive funding rounds — angel, strategic, and Pre-A1 — raising over RMB 500 million in total within just five months of official operations. BlueRun Ventures led its angel round and continued to significantly increase its stake in the strategic round.

As embodied intelligence gains momentum, robots still struggle to precisely perceive force and tactile feedback, adapt to dynamic interaction logic, and perform dexterous operations in advanced scenarios such as precision manufacturing and complex household tasks. The limitations in data fidelity, collection accuracy, and scalable acquisition capabilities have become a core bottleneck constraining industry deployment.

To address these pain points, OriginFlow, a tech company founded by Tsinghua University PhD student Qin Shentao (born after 2000), has developed a new data collection paradigm based on NeuroScale, using neuromuscular sensing technology to break through the physical interaction data bottleneck. Going forward, the company will continue to deepen its R&D in embodied intelligence foundational technologies and accelerate industrial deployment in advanced scenarios.

BlueRun Ventures stated: "We have long maintained close attention to the evolution of interaction technology and its applications across multiple scenarios, actively positioning ourselves for the next generation of interaction entry points. We believe that interaction solutions centered on NeuroScale technology have the potential to become a critical gateway for next-generation human-machine interaction, achieving a value leap from 'interaction peripherals' to a 'general embodied interface layer.' OriginFlow's founding team demonstrates exceptional industrial and technical insight, resource integration and talent attraction capabilities, outstanding execution, and extremely rapid strategic iteration — making them a team we are very optimistic about. We hope they will become industry leaders and breakthrough-makers in the future."

Data quality is the critical gate for embodied intelligence to achieve generalization, and the industry has long been trapped in an "impossible triangle" of cost, quality, and efficiency. Current mainstream technical approaches all have clear adaptation boundaries: first-person visual acquisition solutions represented by EgoScale can only record surface-level behaviors and struggle to reconstruct the underlying logic of human force application; teleoperation and motion-capture gloves face high hardware barriers that limit scalable commercial deployment.

As one of the first companies globally to systematically apply sEMG technology to general embodied intelligence data collection, OriginFlow independently proposed the NeuroScale next-generation collection paradigm. This paradigm uses neuromuscular electrical signals (sEMG) as its core medium, capturing human muscle contraction electrical signals through self-developed EMG acquisition hardware, then algorithmically decoding them to output continuous hand posture, high-resolution force values, tactile feedback, and other multidimensional data. Unlike EgoScale's "observational learning" at the visual level, NeuroScale directly decodes intent and execution logic from the signal source of human movement generation, building an entirely new data foundation for robots to achieve refined, human-like dexterous manipulation.

NeuroScale achieves industry-level breakthroughs across three dimensions:

• Efficiency leadership: By directly interfacing with the human body's native "intent—muscle—action" transmission pathway, it avoids complex transformation processes, significantly improving collection efficiency and shortening training cycles;

• Cost: Through full-stack solutions that optimize hardware structure and integrate supply chain resources, it substantially lowers equipment deployment barriers, making it more suitable for multi-scenario batch deployment;

• Generalization: It possesses strong cross-population, cross-environment, and cross-embodiment generalization capabilities.

Currently, NeuroScale has achieved breakthrough progress in continuous gesture recognition and high-precision force control reconstruction, accurately replicating subtle differences in human force application and effectively addressing industry-wide problems such as rigid robot movements and insufficient force control precision.

Conceptual diagram

After five months of operations, OriginFlow has begun advancing industrial deployment around two major scenarios: industrial manufacturing and household services. Industrial manufacturing scenarios are structured with standardized processes, suitable for high-precision process data accumulation; household service scenarios are unstructured with dynamically changing environments, emphasizing robots' flexible manipulation and generalization capabilities.

In the industrial sector, OriginFlow is partnering with global leading high-end manufacturing enterprises to jointly build embodied intelligence data collection application scenarios. Both parties will focus on high-precision, high-repetition, and high-risk industrial links, using intelligent solutions to optimize production efficiency and improve production line utilization, achieving cost reduction and efficiency gains. This collaboration can effectively compensate for high-end manufacturing labor shortages, avoid human operational errors, replace hazardous, heavy, and repetitive low-end processes, and drive workforce transition toward high-value R&D and management positions. According to industry projections, after large-scale deployment of industrial humanoid robots, short-term industry returns could reach the tens of trillions of dollars, with long-term industrial value potentially exceeding hundreds of trillions.

The household service track possesses vast market space and social value. According to McKinsey & Company research data, global unpaid household and care work totals 680 billion hours annually, with hidden economic value exceeding $12 trillion; if household robots cover 30%-50% of household scenarios, they could generate over $6 trillion in annual industry returns. Beyond this, this household intelligence solution can effectively address the challenge of global population aging. Industry data shows that by 2035, the global population aged 65 and above will exceed 1.2 billion, with a caregiver shortage of over 100 million; after large-scale adoption of household service robots, over $2 trillion in annual operating costs could be saved for elderly care institutions and home-based elderly care scenarios.

In household service scenarios, focusing on high-frequency operational tasks such as clothing organization, home cleaning, and kitchen work, OriginFlow will collaborate with industry partners including 58 Group to collect massive amounts of human operational data, building a skill database for household robots to provide continuous, high-quality "data fuel" for the capability evolution of household service robots, jointly promoting large-scale application deployment of household service robots.

From a longer-term industrial logic perspective, NeuroScale points not only to embodied intelligence data collection but may also extend to next-generation human-machine interaction entry points. As neuromuscular electrical technology matures, low-latency, high-precision, low-intermediary human-machine interaction methods could form new application spaces in robotics, intelligent hardware, medical rehabilitation, spatial computing, and other fields.

Looking at the evolution of human-machine interaction, the PC era relied on keyboard and mouse, the mobile era relied on touchscreens and cameras — each interaction revolution spawned trillion-dollar industrial transformations. Today, OriginFlow is opening a new era of "de-intermediated" human-machine interaction with its globally original NeuroScale paradigm. This technology can not only replace low-end repetitive human labor, drive exponential growth in production efficiency, and create trillion-dollar industrial增量; it will also reshape product forms and business logic in industries such as entertainment, healthcare, intelligent hardware, and electronic information, injecting core momentum into the scaled development of artificial general intelligence.

NaoSi Technology Completes Over $10 Million Angel Round, Focusing on Invasive Brain-Computer Interfaces and Neural Foundation Models | BlueRun Portfolio

BlueRun Ventures Continues to Increase Stake as Zhitai Bio Completes Over RMB 500 Million Series C Round | BlueRun Portfolio

BlueRun Angel Project Kaiwu Ji Completes Hundreds of Millions in Angel+ Round, AI4S Drives Construction of Materials Intelligence Infrastructure | BlueRun Portfolio