Evidence Brief · Workforce Transformation
Changan Automobile: How Domain Experts Build AI Workflows in China
How Changan Automobile's engineers and business employees use AI to build workflows across manufacturing, engineering, quality and procurement in China.
Overview
Changan Automobile offers a practical case of how artificial intelligence is changing work inside China's manufacturing enterprises—not only through centrally deployed automation, but through tools designed by employees who understand everyday operational problems.
Drawing on Changan's 2026 half-year disclosure and attributable September 2026 reporting, this 15-page AIGW Deep Dive examines AI-assisted work across engineering calculations, quality management, vehicle-document checking, supplier packaging, procurement and customer-feedback analysis. It documents reported examples of employees translating domain knowledge into executable workflows, including tools built or configured with QwenWork.
The company's wider AI programme reported 18 AI-master projects, 160 deployed scenarios, more than 3.11 million cumulative employee uses and over 13 million autonomous AI executions. These figures describe the wider programme, not independently verified productivity gains from individual QwenWork applications.
Using AIGW's Workforce Architecture Framework, the report considers how domain experts can become participants in workflow design, while identifying the organizational requirements for verification, reusable capabilities, governance and maintenance.
The evidence does not establish attributable layoffs, net job creation, formal career transitions or independently validated enterprise-wide productivity improvements. The case illustrates emerging mechanisms of employee-led AI implementation and the measurement gaps that remain.
Key messages
- Domain expertise can become AI-building capability. Employees who understand engineering, quality and procurement problems can help define and create AI-enabled tools without necessarily becoming professional software developers.
- AI changes tasks and workflows before formal careers necessarily change. The reported episodes show task compression and workflow redesign, but do not establish company-wide workforce transition outcomes.
- Local experimentation requires organizational infrastructure. Employee-built tools need validation, security, permissions, maintenance, shared knowledge and clearly assigned responsibility to become dependable enterprise capabilities.
- AI adoption metrics are not employment outcomes. Scenario counts, usage figures and task-time estimates cannot establish net job creation, layoffs, higher wages or sustained workforce productivity.
- China's AI manufacturing policies provide context, not causal proof. National and sectoral policy initiatives help explain the implementation environment but do not independently validate Changan's reported results.
About this publication
Suggested citation
AIGW Policy Observatory. (2026). From Domain Experts to AI Workflow Builders: Changan Automobile and the Organisational Work of AI Implementation in China. AIGW Enterprise Workforce Transition Deep Dive. https://gaeedu.org/publication/changan-automobile-ai-workflow-builders-china
References
Related Workforce Evidence
Changan Automobile
Engineers build an AI-assisted harness-selection workflow with technical verification
China
Mao Hai formalized harness rules and used AI to build an executable tool; structured inputs replace manual calculation, with engineer checking.
Related Publications
Evidence Brief
LONGi: How AI Quality Inspection Is Turning Frontline Expertise into AI Training
How LONGi's Jiaxing solar factory uses AI quality inspection, with documented inspector-to-AI-trainer pathways, human verification and evidence limits on labour reduction.
October 2026
Flagship Report
Workforce Architecture for the AI Enterprise
Drawing on nine enterprise AI transitions, this flagship report examines how organizations redesign workflows, roles, capabilities, human authority and careers as AI moves from adoption into everyday operations.
12 October 2026