Lingyan Technology, a Hangzhou-based startup developing bionic dexterous faces and emotional interaction systems for robots, has completed tens of millions of yuan in combined seed and seed+ financing. Seed-round investors include In Capital and Zhejiang Torch Center; seed+ investors include Yida Capital, Plum Ventures and Linglu Capital, with existing shareholder In Capital making an additional investment. Linglu Capital served as the financial advisor for this and subsequent rounds. The proceeds will be used for product iteration, core component R&D, algorithm platform construction and team expansion.
Bionic dexterous face full-stack product: FaceSoul Agent
Lingyan Technology was founded in June 2026 and is headquartered in Hangzhou. Its core product, FaceSoul Agent, is described by the company as a full-stack technical architecture for bionic dexterous face agents that synchronizes audio, lip and eye movements to provide an emotional interaction solution for humanoid robots, service robots and companion devices.
The company defines FaceSoul Agent as composed of two parts: the “face,” responsible for high-fidelity physical presentation through bionic skin, muscle transmission design, expression temperature and posture perception; and the “soul,” which carries large AI models, social cognition, memory learning and an emotional engine to provide self-understanding, emotional cognition and value judgment.
Product lines launched by Lingyan include Lingshu M1 for education and research, Lingshu W1 for commercial services, the consumer-oriented Lingyun Series and the Lingneng Series for humanoid robot OEMs. The company also offers IP customization services.
Bionic dexterous face technology and engineering choices
Lingyan’s technical path emphasizes cable-driven actuation for facial motion rather than the more common link-drive approach. The company says cable-driven design reduces noise and spatial occupation and better imitates human muscle arrangement, enabling more anthropomorphic and delicate expressions. Its current products reach 32 degrees of freedom in the face, claimed as the highest among similar domestic products.
At the software level, Lingyan focuses on expression-generation algorithms rather than competing directly in large dialogue models. The company developed lightweight end-side expression generation models using cross-modal distillation, continuous collaborative modeling and end-side quantitative deployment to close the loop from speech understanding to expression execution. Lingyan positions this capability as a solution for OEMs that typically handle dialogue and memory themselves but require real-time, synchronized audio-to-lip-and-eye expression conversion.
Founder and CEO Zhao Changyi, a post-2000s PhD from Jilin University, credits technical reuse across an ecosystem formed with fellow alumni and companies originating from the Key Laboratory of Bionic Engineering at Jilin University. Zhao said the cable-driven path derives from experience at Heiman Technology in dexterous hands and micro drive motor modules reuse from QuanzhiBo in robot joints. On the technical choice, he stated that “we are the only domestic enterprise that is seriously doing cable-driven dexterous face, because cable drive requires long-term accumulation.”
Zhao also argued on costs: “On the contrary, the cost of cable drive is even lower, because there are too many non-standard parts for links,” and explained that cable drive better imitates muscle driving where required forces are small.
Manufacturing, product roadmap and market strategy
Lingyan says it has adopted a modular hardware design with independent eyebrow, eye and neck modules intended to enable production across four separate lines. In early stages, the company expects to leverage production capacity from partner firms Heiman and QuanzhiBo if its own lines cannot meet demand.
Within four months of establishment, Lingyan reported cooperation or intended cooperation with operators, cultural tourism exhibitions, commercial service providers and some humanoid robot OEMs. The company plans to focus this year on direct To B/To C sales and IP customization, which it expects to account for about 90% of revenue; it projects that next year, OEM supporting business will account for roughly half of total revenue as OEM demand ramps up.
On product iteration, Lingyan plans a future full cable-driven product with 43 degrees of freedom (46 including the neck) to “fully realize human-like presentation.” The company is also pursuing overseas opportunities: a Middle East customer interest in health monitoring has prompted promotion of a “Little Ling Doctor” product line, while IP customization will be promoted in Europe and the U.S.
On the uncanny valley, Zhao said he considers 75%–85% anthropomorphism to be the interval where the effect is strongest; Lingyan aims to keep C-end and B-end product anthropomorphism above 90%, and to control IP-customized designs (animals, cartoons) below 65% to avoid the uncanny valley.
Investors’ rationale
Seed+ investor Yida Capital described the emotional expression layer as a largely undeveloped, high-value segment of the embodied intelligence industrial chain and highlighted the synergies between joints, dexterous hands and dexterous faces in the Jilin University-derived ecosystem. In Capital, a seed investor that increased its stake, praised the team’s rapid progression from concept verification to product finalization and emphasized the company’s real-time expression generation without reliance on preset libraries. Plum Ventures cited exclusive synchronized face data and real-time, model-based expression generation without preset libraries as key assets. Linglu Capital said Lingyan’s large-model-based real-time expression generation and scene adaptability have no similar solutions in China and pledged lifecycle resource support. Zhejiang Torch Center framed Lingyan as an example of translating laboratory research into industrialization and pledged continued policy and incubation support.
The funding and investor backing position Lingyan to pursue product iteration, core-component development, algorithm platform building and team expansion as it seeks to commercialize bionic dexterous face technology in domestic and selected overseas markets.
