Generative AI Integration for PLM.

OVERVIEW

The challenge behind the work.

Designed and developed a Generative AI solution integrated with Product Lifecycle Management (PLM) systems, enabling users to query PLM information using conversational language instead of navigating complex data structures and reports.

Make product structures, documents, metadata and project information easier to discover through natural-language interaction while grounding responses in enterprise data and respecting user permissions.

APPROACH

Architecture & approach

1. 01Connect Large Language Models (LLMs) with PLM repositories and PostgreSQL databases.
2. 02Use Retrieval-Augmented Generation (RAG) to ground responses in enterprise information.
3. 03Convert natural-language queries into SQL and PLM-specific searches.
4. 04Index documents and support semantic retrieval with embedding models and vector databases.
5. 05Integrate local LLMs through Ollama for secure, on-premises deployment scenarios.
6. 06Enforce access controls so responses respect existing user permissions.

CONTRIBUTION

Technical contributions

• Developed an AI-powered assistant for PLM data retrieval and knowledge discovery.
• Integrated LLMs with PLM repositories and PostgreSQL databases.
• Implemented RAG-based responses, natural-language-to-SQL conversion and PLM-specific search.
• Built document indexing and semantic search capabilities using embeddings and vector databases.
• Integrated Ollama local LLMs to support secure on-premises deployments.
• Implemented access controls to align AI responses with existing user permissions.