AI IntegrationRAG / Knowledge System4-month engagement
RAG-Based Engineering Knowledge Assistant Connected to Windchill Document Library
An industrial OEM had 12 years of engineering documentation in Windchill that nobody could search. Engineers spent hours looking for past decisions and often redid work that already existed.
المشكلة
Windchill full-text search returned only metadata matches. The actual content of 600K+ PDF and Office documents was inaccessible. Engineers queried colleagues instead of the system, creating knowledge bottlenecks around senior staff.
الحل
Built a RAG pipeline that indexes Windchill document content into a vector database via the Windchill REST API and Apache Tika for document parsing. Engineers query in natural language and receive answers with citations linking back to source documents in Windchill.
التقنية
Windchill REST API, Apache Tika, OpenAI GPT-4o, text-embedding-3-large, Pinecone vector database, FastAPI backend, React frontend, SAML SSO integration
600K+
Documents indexed
45m→5m
Average document search time
92%
Answer relevance score (UAT)
4 mo
Proposal to production
تعمل على شيء مماثل؟
أخبرنا بما لديك. خلال 30 دقيقة يمكننا أن نخبرك إن كان مناسباً وما الذي سيتطلبه.