Xpeng Motors (小鹏汽车)

Solution

Xpeng built a company-wide digital employee platform on Alibaba Cloud to unify knowledge search, content generation, workflow assistance, and management insight under one governed AI layer. The platform blends RAG over FAQs, internal documents, external knowledge, and open search with knowledge-graph enrichment and multi-path retrieval so employees can get concise answers without manually navigating many systems. At the model layer, Xpeng uses a hybrid architecture that routes sensitive or high-permission requests to privately deployed models on PAI while AI Gateway can burst overflow traffic to Bailian during peaks and apply policy across both environments. AI safety guardrails add HTTPS transport, content filtering, and intent classification so internal questions that touch private knowledge stay within the private domain. On top of that base, digital employees cover help-desk Q&A, chatbots, document generation, multimodal creation, training, task handling, and management analytics.

Data flow

Employee questions, search requests, and generation tasks enter the digital employee layer, which retrieves context from internal and external knowledge sources through RAG and knowledge-graph enrichment. API Gateway then routes the request either to private models on PAI or to Bailian burst capacity, while AI safety controls filter content and restrict sensitive data to the private environment before answers or generated assets are returned.