We are Beijing WeLinkirt DaoAI Technology Co., Ltd. (brand name DaoAI; overseas entity DaoAI Robotics Inc.), founded in 2018 in Vancouver, Canada and Beijing, China. We are a leading enterprise in sovereign physical AI and a pioneer of robot cognition technology, and we define how physical AI is deployed in industry: we are the only physical-AI company in the world to have built a four-layer full-stack moat in industry — foundation model, 3D camera, vertical agent and the DaoAI ACI cognitive inspection robot. We specialise in physical artificial intelligence and are recognised as a national key "Little Giant" specialised and sophisticated enterprise. We maintain our operating headquarters and research centre in Beijing, regional offices in Wuhu (Anhui) and Chongqing, a global manufacturing base under construction in Jiangjin (Chongqing), and an embodied-intelligence research unit in Vancouver.
Our founding team brings together a North American computer-vision research group and alumni of Peking University and Tsinghua University; our chief scientist is a former principal AI researcher at Siemens.
I. Technology position
We hold that artificial general intelligence consists of one general world model together with a number of vertical small models, rather than a large language model extended with tool calls and agents. Language models excel at composing symbols and knowledge; manufacturing, robotics and video surveillance instead turn on how objects exist and move in three-dimensional space, and why they move as they do. Those constraints are not present in text corpora; they exist only in the physical world.
On this basis we have independently developed DaoAI World, a visual foundation model that jointly represents space, material, light, motion and causality, and from which task-specific models for each scenario are derived. The foundation model provides a unified representation of the physical world; the derived models are responsible for accuracy and latency in each scenario.
II. Product lines
On our proprietary DaoAI World physical world model we upgrade the traditional "AOI optical inspection" into a new category, ACI (Auto Cognitive Inspection). Three layers extend downwards from that foundation: industrial structured-light 3D cameras (global manufacturing base in Jiangjin, Chongqing); the DaoAI ACI cognitive inspection robot operating system — an "inspection brain" driven by the feature cognition of a visual foundation model, deployable via SDK, API or Docker and embeddable in third-party 2D or 3D equipment, running entirely on premises; and the DaoAI ACI cognitive inspection robots themselves — 2D and 3D systems in more than ten models, covering inline, offline and insertion stations.
Four product lines are built on DaoAI World, sharing one foundation model, one self-training platform and one compute-scheduling layer:
- Industrial ACI (AI-AOI) inspection — an ACI OS, 2D ACI systems and 3D ACI systems, covering full-board SMT and PCBA appearance inspection as well as three-dimensional height and coplanarity inspection.
- Robotic vision — 3D vision-guided random bin picking, machine tending, dispensing and assembly, with six-degree-of-freedom pose estimation and sub-millimetre hand-eye coordination.
- SkyVision intelligent surveillance — a zero-code video surveillance AI platform on which customers train their own models on site within hours, with data remaining entirely on premises.
- Wemio enterprise content platform — built for corporate marketing departments, covering corporate IP, brand communication, founder IP, product introductions and corporate story dramas. Final rendering runs on Wemio Aster, our proprietary model, in which genre, colour, lighting, camera movement and camera position are all adjustable parameters.
III. Sovereign AI
In response to data-localisation requirements in critical information infrastructure sectors such as government, transport and energy, we build our capability across three layers: data sovereignty, technology sovereignty and operational sovereignty. Training data and production records remain entirely within the customer's own facilities; we develop the full stack in house, from sensors and three-dimensional imaging through to the world model inference engine, without dependence on any external large-model API; and integration is offered via SDK, API and Docker, leaving deployment form and schedule to the customer.
IV. Core capabilities
Our products are based on the feature cognition of a visual foundation model: a single good sample is sufficient to build a model, which is automatically programmed and brought online in approximately five minutes, without CAD drawings and without a pre-collected defect library. Product changeover requires no code, and models continue to improve from production-line feedback. Our proprietary 3D industrial cameras use structured light with HDR imaging, cover a field of view from 77 to 477 mm, achieve 4 μm repeatability, complete 3D reconstruction in 0.8 seconds, are calibrated on site within three minutes, and operate continuously around the clock; point-cloud accuracy reaches 1 μm (a typical value under laboratory calibration; this is a different metric from the repeatability quoted above, and on-site measurement governs).
On lines already in production, DaoAI's ACI systems reach an inspection accuracy of 98% or better from one-click modelling, and false calls drop by roughly 80% against an existing rule-based system. Inference runs at the edge of the line, with a decision latency below 12 ms; everything runs offline, data never leaves the plant, and the models keep sharpening on production-line feedback. (typical achievable ranges in anonymised industry scenarios; actual figures vary with process, component and existing defect library, and on-site measurement governs)
V. Customers and ecosystem
We have signed contracts with Siemens, Brose, Adient, Midea, AUX, Inventec, Hebei Expressway, PowerChina and China Mobile, among others, and conducts ecosystem cooperation and pilot projects with Foxconn, BYD, CATL, Luxshare and Huaqin. We are an NVIDIA ecosystem partner and a member of Canada's NGen "AI for Manufacturing" programme, maintain strategic cooperation with Inspur and Unisplendour Xiaotong, and list our products on the Inspur Yuannao AIStore.
VI. Credentials
DaoAI is a national specialised-and-sophisticated "Little Giant" enterprise and a central-government-funded key "Little Giant" enterprise (reviewed by the Ministry of Industry and Information Technology), and has been admitted as a working-group member of the MIIT standardisation technical committee for humanoid robots and embodied intelligence. Our company and products have been the subject of several Xinhuanet features. We currently hold 32 granted patents, of which 15 are invention patents, together with 96 software copyrights, and have been selected as a typical artificial-intelligence application case by the Ministry of Industry and Information Technology (one of 285). In July 2026 we completed our Series B financing, led by the Chongqing Industry Guidance Fund.
VII. Why it matters
High-end industrial inspection and machine vision have long been led by overseas suppliers such as Koh Young (Korea), Omron (Japan) and Cognex (United States), mostly under a rule-based AOI paradigm that programs board by board from handwritten rules and large defect libraries. With a four-layer, domestically developed ACI stack, DaoAI offers a substitute that goes beyond that paradigm, fills a gap in high-end domestic inspection equipment and gives the electronics manufacturing supply chain a local option — in line with national priorities around "AI Plus", embodied intelligence, manufacturing strength and data security.