AI knowledge base: 12K docs, 40s to retrieve
12K process, quality and compliance docs turned into a cited, auditable engineering assistant.
AI knowledge base
Advanced manufacturing · 鼎力智造 · 2025
Dingli Manufacturing keeps a large volume of SOPs, work instructions, quality standards and compliance docs scattered across PLM, document management, quality and ERP. A new engineer spends an average of tens of minutes finding a single standard, and practices diverge between plants.
External audits (ISO 9001 / IATF 16949) consume huge effort every year: a large portion of auditor questions require digging through files, and a stopped line is a real cost. We needed an auditable, versioned, role-isolated AI knowledge base that fuses find-doc → find-passage → cite-source in one move.
How Ouryun ships AI knowledge base for 鼎力智造
RAG knowledge base at the core, AI Gateway as the front door, a model platform that separates offline parsing from online inference, governance that enforces citation and role isolation, and apps covering web search, WeCom chat and on-floor displays.
RAG knowledge base
12K documents are chunked, versioned and indexed, isolated by process, plant and product line, with monthly incremental sync.
AI Gateway
Floor terminals, office PCs, WeCom and PLM connect through one gateway with role-based access (engineer / auditor / operator).
Model platform
Offline document parsing + online inference — vision models parse scans and drawings offline, a domestic LLM handles semantic retrieval online.
Citation-first answers
Every answer cites the original passage with a clickable link to the source PDF; sensitive queries are logged automatically.
Three surfaces, one experience
Web search, WeCom chat and on-floor displays — engineers, auditors and shift leads all have an entry point.
From connect to ship, in four steps
Connect
In 2 weeks we connect PLM, document management, quality and compliance through API / file sync, and map document lineage.
Govern
In 3 weeks we ship version control, role / permission matrix, citation provenance, and confidential-tiering policies.
Deploy
In 4 weeks we ship web search, WeCom chat and on-floor displays to 100 seed engineers, then iterate twice.
Scale
In 8 weeks we roll out to 6 plants, run monthly incremental sync and quarterly full retraining, and push on-floor hit rate above 92%.
Process, quality and compliance docs behind RAG — engineers get cited answers on the floor.Full case study delivered and reviewed by the Ouryun team.
Numbers that prove value
Bring this capability into your business
More Ouryun case studies
Manufacturing AI quality inspection
An AOI + AI inspection appliance for PCB production that performs AI-based second-pass review on AOI NG images and reduces manual review load.
Domestic AI inference resource pool
A unified domestic AI inference resource pool for a large manufacturing group, supporting multiple large models and group-level AI platform operations.