Artificial intelligence is developing into a decisive competitive factor for small and medium-sized enterprises (SMEs). However, practical implementation frequently fails due to technical challenges and high investment costs. The LLM4KMU research project addresses these hurdles through the optimization and simplification of Open Source Large Language Models (LLM) deployment in SMEs.
The project is funded under the NEXT.IN.NRW program of the state of North Rhine-Westphalia and creates new opportunities for the digital transformation of SMEs.
Objectives and Technical Approach
The continuous development of large language models, known as Generative AI, significantly expands the technological possibilities for companies. SMEs can optimize their processes through intelligent automation, increase efficiency, and reduce costs. LLM4KMU develops customized solutions for integrating these technologies and enables companies to access AI applications without extensive investments or complex technical implementations.
Development Environment for Industrial AI Integration
The core of the project is a professional development environment where companies can systematically evaluate various language models and identify the one that optimally fits their industrial requirements. The focus is on open-source models – freely available AI models that can be further developed and adapted by a broad developer community.
The technical approach is based on a modular integration concept: Language models can be systematically integrated, evaluated, and replaced with more powerful alternatives when needed – without extensive technical adjustments. This flexibility enables companies to continuously optimize their AI solutions and adapt to changing requirements, generating sustainable value creation. The goal is to develop concrete, production-ready AI applications for industrial practice.
Data Protection and Best Practices
A central aspect of the project is the development of comprehensive best practice guidelines that support companies in optimizing and adapting their LLM. This leads to more reliable models and reduced usage costs. Particularly important: Companies retain complete control over their data, as all solutions are developed on an open-source basis and can be seamlessly integrated into existing IT infrastructures.
Strong Consortium of Science and Industry
The project is supported by a high-performance consortium of science and industry. In addition to the University of Bielefeld as coordinator, semalytix GmbH, CLAAS Selbstfahrende Erntemaschinen GmbH, ellamind GmbH, ZENIT GmbH, primeLine Solutions GmbH and Matplus GmbH are involved. Each partner contributes their specific know-how to facilitate companies’ entry into the AI world and strengthen the innovation power and competitiveness of North Rhine-Westphalia.
