Brückner Maschinenbau, a market leader in film stretching lines for packaging and technical films (BOPP, BOPET, battery separators), faced a critical operational challenge: an ever-growing database of over 500,000 parts where maintaining data consistency had become increasingly laborious. Material master data ingested from external sources — supplier libraries, certificates, specifications — was incomplete, inconsistent, and heterogeneous; duplicates, varied nomenclature, and missing change logs made data management error-prone and costly. Skilled Brückner engineers often spent valuable time searching for the right materials, or missed critical change notices for components, due to duplicates in the database.
Solution: Consulting partner Zühlke designed and implemented a bespoke GenAI-powered application using customised Large Language Models (LLMs) to automate data quality. The application enables data managers to upload data, correct it with LLM-assisted suggestions, then reintegrate the refined results into the master database. Continuous technical and user acceptance testing ensured smooth integration into Brückner's workflow; the Minimum Viable Product (MVP) achieved highly promising levels of accuracy.
Tech details: Customised LLMs for automated data harmonisation; bespoke user-facing app for upload → LLM-assisted correction → reintegration; iterative development with technical and UAT cycles; user interviews and data exploration as foundation. The same partnership subsequently extended to predictive AI for packaging film quality — explainable AI models capture key underlying physical mechanisms to predict film quality before production from machine and material parameters (dozens of material types, hundreds of machine parameters), reducing waste and improving overall efficiency. These insights will serve as the foundation for AI-based predictive assistants developed in the future.
Metrics: 500,000+ parts in master data database; MVP saved several hours per week for data managers; hundreds of data entries corrected within the first week of use; AI models extended to dozens of material types and hundreds of machine parameters for film quality prediction; AI-based predictive assistants in development for future production efficiency.
Quote: "The collaboration with Zühlke was a natural choice, given their proven track record in AI and industry-specific solutions." — Brückner
"Zühlke's role was pivotal. The team combined strong industry know-how, deep expertise in applied AI, and a user-centric approach to support our journey from strategy to implementation." — Brückner
Closing: The successful LLM deployment sparked a wave of enthusiasm within Brückner and opened doors to further AI initiatives across the company. The MVP was delivered in record time, and the partnership continues with AI-based predictive quality assistants and additional production efficiency applications. Article co-authored by Thomas Grünäugl, Engineering Electric - Material Planning, Brückner Maschinenbau, with Zühlke.