On 24 September Beijing Dahu Zhiyuan Technology closed a tens-of-millions-yuan Pre-A round led solely by Eagle Fund, a vote that its model of rebuilding intangible-cultural-heritage food through embodied intelligence has market validation.
Most service-robot projects chase back-kitchen cost savings. Dahu Zhiyuan took a different path, anchoring its technology to the living heritage of Chinese snacks, from Jianbing pancakes and traditional noodles to wok-fried dishes, and offering the sector a differentiated commercial reference.
Why heritage food, not kitchen labour
Many automation plans start from easing store staffing pressure. Dahu targets a harder problem: the transmission gap in food heritage. Hundreds of years of local snacks live only in master cooks’ muscle memory, with heat, force and recipe ratios undocumented and passed by word of mouth. As masters age, whole flavour systems risk vanishing.
The core team brings smart-hardware and machine-vision experience. Early on they realised that typing a paper recipe into a machine cannot recover the texture unique to heritage food. The founder judges that embodied intelligence’s value reaches beyond replacing labour: it can distil the tacit, hard-to-verbalise craft in a master’s head into a saveable, copyable digital asset.
Eagle Fund backed the project at an early stage not for the hardware but for the chance that technology could reach everyday street-consumer scenes. Many AI projects stall in lab demos. Dahu planted its technology in ordinary snacks and tried to solve heritage food’s resistance to chain expansion.
How the robot copies the master’s hand
Most food robots are fixed-sequence devices. When humidity, ingredient state or temperature drift, taste wavers, an industry pattern called blind execution. Reproducing pancakes or hand-pulled noodles demands real-time perception of material state and dynamic adjustment of the whole cooking flow.
Dahu built its own dynamic algorithm framework fusing machine vision, multi-dimensional perception and adaptive temperature control. The team records the full artisan process and breaks down heat changes, rhythm and dosing logic into data models an AI can run, then adjusts parameters within milliseconds to keep flavour stable and cut reliance on individual masters. Against fixed-timer machines it absorbs variation from environment and ingredients.
The integrated unit needs no complex kitchen rebuild and fits tourist sites, malls, campuses, transit stations and corporate sites with round-the-clock output. The company already has batch production capacity, the base for wider rollout.
Can capital fix the hard parts
The fresh capital flows to deeper AI algorithms, more heritage-food digital models and a broader product matrix, plus nationwide expansion and cross-domain talent. Inside the firm there is agreement that scaling heritage food needs three layers: digitised ancient craft, hardware that runs steadily across scenes, and a team mixing research, cooking and cultural operation.
The round also signals a shift in how investors judge service robots, once measured mainly on hardware specs with weak commercial loops. Dahu’s blend of technology, heritage, tourism and retail sells a full digital-operation solution, not a single device. Embodied-AI commercialisation is not limited to costly humanoids. Packaging AI into a vertical consumer terminal and a narrow scene also works. The machine can stabilise flavour and fix process, yet it cannot fully reproduce a maker’s human warmth, and technology should assist heritage rather than erase the artisan.
Editor’s note: This is an adapted translation of the original OFweek report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://robot.ofweek.com/2026-09/ART-8321203-8420-30704743.html.