Incubated by Liu Chunbao of Jilin University and founded by a post-00s team, Lingyan Technology's bionic dexterous face completes tens of millions of yuan in financing

36Kr has learned that Hangzhou Lingyan Technology Co., Ltd. (hereinafter referred to as "Lingyan Technology"), a company specializing in biomimetic dexterous faces and emotional interaction technology, recently completed a seed round and seed+ round of financing totaling tens of millions of yuan. Seed round investors include Yingdong Capital and Zhejiang Torch Center; seed+ round investors include Yida Capital, Plum Ventures, and Linglu Capital, with existing shareholder Yingdong Capital continuing to increase its investment. Linglu Capital served as financial advisor for this and subsequent rounds. The funds will be mainly used for product iteration, core component R&D, algorithm platform construction, and team expansion.

At a time when the embodied intelligence and humanoid robot industries are advancing by leaps and bounds, while most attention is focused on robots' "limbs" and "torsos," Lingyan Technology has chosen a highly vertical yet critical track—making the "faces" of humanoid robots.

Lingyan Technology was founded in June 2026 and is headquartered in Hangzhou. It is China's first company to achieve synchronized audio, lips, and eyes in biomimetic dexterous face intelligent agents, providing the FaceSoul Agent full-stack emotional interaction solution for humanoid robots, service robots, and emotional companion devices.

The company's founder and CEO, Zhao Changyi, is a post-2000s PhD from Jilin University. He was a core member in the early stages of Wuxi Quanzhibo (a leading domestic joint motor company) and Hangzhou Heiman Technology (a dexterous hand supplier to Leju Robotics). All three companies originated from the Engineering Bionics Key Laboratory of the Ministry of Education at Jilin University, and their founders all studied under Liu Chunbao, a leading professor and "Tang Aoqing Scholar" (who currently serves as chief scientist of the above three companies), forming close collaborative relationships across the industry chain.

Relying on this industrial ecosystem, Lingyan Technology has launched its core product—FaceSoul Agent. This is a full-stack biomimetic dexterous face intelligent agent technology architecture, designed to provide humanoid robots, service robots, and emotional companion devices with a complete emotional interaction solution from "form" to "soul."

In Lingyan Technology's definition, FaceSoul Agent consists of two parts. "Face" (biomimetic form) is responsible for high-fidelity physical presentation, restoring the most realistic human facial forms and reactions through biomimetic skin, muscle transmission design, expression temperature, and posture perception; "Soul" (the core of the soul) carries AI large models, social cognition, memory learning, and an emotion engine, endowing robots with self-understanding, emotional cognition, and value judgment capabilities.

In terms of product matrix, the company has launched four major series: the "Lingshu M1" for the education and research market, the "Lingshu W1" for commercial services, the "Lingyun series" for the consumer market, and the "Lingneng series" for humanoid robot OEMs. In addition, the company also provides IP customization services.

Product display (Image source/company)

The background of founding by fellow students has brought Lingyan Technology an extremely high technology reuse rate and cost advantages. Zhao Changyi told 36Kr frankly that Lingyan Technology's tendon-driven technology path comes precisely from Heiman Technology's accumulation in the dexterous hand field, while the micro drive motor module technology is reused from Quanzhibo's accumulation in robot joints.

What key role does tendon-driven technology play in humanoid robots' "faces"? At present, most biomimetic face companies on the market use linkage drive, but in Zhao Changyi's view, linkage drive has pain points such as high noise and large space occupation, and the quick-return characteristic of linkages produces sudden speed changes at the end.

Lingyan Technology is one of the few domestic companies taking the tendon-driven dexterous face technology path. Tendon-driven technology not only saves facial space, enabling the product's facial degrees of freedom to reach 32, the highest among similar domestic products, but also makes expression generation more human-like and delicate by imitating the arrangement of human muscles.

In addition, at the software and algorithm level, Lingyan Technology has avoided the fierce competition of conversational large models and firmly focused its R&D on "expression generation algorithms."

Zhao Changyi believes that OEMs usually handle functions such as dialogue and memory themselves, while their rigid demand for dexterous face suppliers is "to convert voice and visual input into speech-capable, precisely aligned mouth shapes and eye movements." This is a process of encoding phonemes, prosody, semantic states, and emotional intentions in Chinese speech into a unified temporal representation and generating continuous expressions in the robot's multi-degree-of-freedom motion space.

In addition, there is another core pain point in the industry: most peers' computing power relies on external local servers and is difficult to truly integrate into the humanoid robot body.

Relying on a large amount of high-quality facial data and bionic project experience from the Jilin University laboratory, Lingyan Technology has self-developed a complete set of lightweight on-device expression generation models. Through cross-modal distillation, continuous collaborative modeling, and on-device quantized deployment, it has closed the loop from speech understanding and expression generation to execution feedback.

Speaking of performance and implementation, Zhao Changyi revealed that in the four months since its founding, the company has reached cooperation or intended cooperation with telecom operators, cultural tourism exhibitions, commercial services, and some humanoid robot OEMs. The company plans to focus this year on To B/To C direct sales and IP customization, accounting for about 90% of total revenue; next year, as OEM demand scales up, humanoid robot supporting business will account for half.

Regarding the "uncanny valley effect" when robots interact with humans, Zhao Changyi has a clear understanding: 75%-85% human-likeness is the uncanny valley range, and beyond 85% it no longer applies.

Therefore, through the texture of biomimetic skin materials and high-precision algorithms, Lingyan Technology has achieved over 90% human-likeness in its C-end and B-end product lines; while in the IP customization line (such as animals and cartoon characters), it deliberately controls human-likeness below 65% to avoid the uncanny valley range.

The following is an excerpt from the conversation between 36Kr and the company's founder and CEO Zhao Changyi (edited):

36Kr: Currently, everyone in the industry generally uses linkage drive. You insist on the tendon-driven route—will there be cost pressure? Where will the industry go in the future?

Zhao Changyi: Actually, it's the opposite—tendon drive is actually a bit lower in cost, because there are simply too many non-standard parts in linkages. The reason everyone uses linkages now is that many companies have recently rushed into the expressive face track, and everyone is racing against time to quickly reach the first tier, so they first use the linkage form, which has a shorter design cycle.

But in the future, faces will definitely converge on the tendon-driven path. Hands are hard to converge, because tendon drive has fatigue life issues; but the force with which a face pulls the skin is very small, and it is precisely this tendon-driven pulling form that can mimic the driving mechanism of muscles. Currently, we are the only company in China seriously using tendon drive, because tendon drive requires long-term accumulation—for example, the position of each muscle and the winding form of each tendon—and we have done biomimetic mechanism tests for all of these in the lab.

36Kr: In hardware mass production and implementation, what advantages does Lingyan Technology have?

Zhao Changyi: At the hardware level, the biggest pain point in the industry right now is that no company can mass-produce; most production lines can only be assembled by hand. But Lingyan has adopted a modular design, with the eyebrows, eyes, and neck each forming separate modules, so in the future we can directly go to four production lines to manufacture them separately.

As for capacity assurance, a foreign partner recently raised a demand for several hundred units. We have one advantage: in the early stage, if our own production lines can't handle it, we can first use Heiman's or Quanzhibo's production lines to help us fulfill orders.

36Kr: Where is the explosive inflection point for the expressive face market? Besides B-end and C-end, what other growth areas are in your future business plan?

Zhao Changyi: I judge that there will be a concentrated outbreak at the end of this year or in Q1 and Q2 of next year. Because the application of humanoid robots in factories is gradually converging, and OEMs are all shifting toward To B and To C scenarios; they lack the ability to make faces and need to find external suppliers.

This year's business will mainly be self-operated products, with IP customization and OEM supporting business as supplements; next year we will focus on increasing the weight of OEM supporting business, making it an important support, while continuing to advance direct-sales B-end and IP customization businesses, forming diversified coordinated development.

In addition, we are now working on overseas markets simultaneously. For example, customers in the Middle East are very interested in health monitoring, and we are promoting the "Xiaoling Doctor" product line; in Europe and the US markets, we will advance IP customization sales. On product iteration, in the future we will make a fully tendon-driven product with 43 degrees of freedom (46 degrees of freedom including the neck), fully achieving human-like presentation.

Investor perspectives:

Seed+ round investor Yida Capital: Humanoid robots are crossing from "technical validation" into the deep waters of "large-scale commercialization." Competition in the body and motion control segments has become increasingly full, and the emotional expression layer is a value highland that is almost blank in the industry chain. We have successively invested in the integrated joint company Quanzhibo and the dexterous hand company Heiman Technology, and have built long-term trust with this team originating from the Key Laboratory of Bionic Engineering of the Ministry of Education at Jilin University. Joints, dexterous hands, and expressive faces share supply chains, customers, and industrial resources, empowering each other in technology and manufacturing, forming an ecosystem approach. Lingyan has fully self-developed everything from bionic skin and micro-drives to emotion models, with zero external dependence, constituting a competitive barrier that is difficult to replicate. Investing in Lingyan is investing in the infrastructure of the embodied intelligence interaction entry point.

Seed and Seed+ round investor Yingdong Capital: We believe robots should not only have a torso and limbs, but also a face that can express. The expressive face is the most scarce interactive component in the embodied intelligence industry chain. Lingyan's real-time expression generation does not rely on preset libraries, and its human-like interaction with synchronized voice, lips, and eyes has reached a leading level in China. After the seed round, we chose to continue adding investment because the team completed the leap from concept validation to product finalization in an extremely short time, and its productization capability and engineering efficiency exceeded our expectations. Yingdong Capital is firmly optimistic about teams that can truly turn technology into products, and is willing to accompany them further with patient capital.

Seed+ round investor Meihua Venture Capital: Early-stage investment is about investing in future scarcity. The first half of embodied intelligence competes on motion control, and the second half competes on product implementation and customers paying. The expressive face is precisely the interaction entry point for robots moving from "able to move" to "able to communicate." Meihua Venture Capital has long focused on entrepreneurship by Jilin University alumni: behind Lingyan is the only ministerial-level key laboratory in China with bionics as its core research direction, and its exclusive synchronized face data constitutes a data barrier that cannot be bought even with money. Technology can be bought, and large models can be called, but such a team and platform cannot be bought. Lingyan's expressions are generated in real time by an emotion foundation model and do not rely on preset libraries; its combination of synchronized voice, lips, and eyes with dynamic pupils has no comparable product in China. Upstream sells shovels, and hardware is king; we are optimistic that Lingyan will become the definer of the emotional expression layer.

Seed+ round investor Linglu Capital: Linglu Capital focuses on frontier technology tracks such as AI and embodied intelligence, and is committed to discovering and long-term empowering future unicorn companies. Every leap in computing platforms is essentially a leap in interaction interfaces. The interaction interface of the robot era is not a screen, but an expressive face. Lingyan's real-time expression generation technology based on large models has no comparable solution in China, and its product form and scenario fit lead the industry. We are optimistic about Lingyan's scarce positioning in the emotional expression layer and the team's engineering speed, and will provide full-lifecycle resource empowerment around the company, accompanying it to become a key participant in embodied intelligence interaction components.

Seed round investor Zhejiang Torch Center: Lingyan Technology is a typical target of "investing early, investing in hard tech, and investing in domestic substitution." The team originates from the Key Laboratory of Bionic Engineering of the Ministry of Education, has dual attributes of artificial intelligence and high-end equipment, and possesses a complete capability chain from basic research and core components to full-machine mass production. It is a vivid example of scientific and technological achievements moving from the laboratory to industrialization—the "last mile." The center has been deeply engaged in science and technology services for more than 30 years and will continue to provide achievement transformation, policy matching, and entrepreneurial incubation support for such local technology enterprises that master core technologies and are at a key stage of industrialization.