AI-driven innovations in research and practice
Generative AI (GenAI) and large language models (LLMs) are far more than just trends – they are revolutionizing business processes through automation, enabling personalized customer experiences, and supporting data-driven decision-making. At the same time, they enhance human creativity and boost efficiency across a wide range of applications. fortiss actively advances these technologies, shaping their ongoing development in terms of reliability, usability, and compliance – for science, industry, and society.
In close collaboration with industry partners, fortiss develops practical AI solutions that enhance decision-making and optimize operational workflows. Our focus areas include Natural Language Processing (NLP), knowledge graphs, and edge-cloud orchestration. Through seamless integration into existing software systems, we enable scalable, secure architectures that meet the highest quality standards – while simultaneously reducing unpredictable system behavior.
fortiss integrates large language models (LLMs) into software and systems development to optimize development processes and modeling. This approach enables the automated generation of modeling artifacts from natural language, the transformation of requirements into functional architectures, and the enhancement of workflow efficiency.
Furthermore, fortiss applies generative artificial intelligence (GenAI) in software development to improve software quality through advanced code generation techniques. In the field of automotive engineering, GenAI is used to generate image data for validation tests of autonomous vehicles.
Our scientists develop advanced technologies to optimize the quality of LLM outputs. This includes improving text normalization, enhancing output quality, and automating evaluation. Methods such as LLM2Vec for data organization and the "LLM-as-a-judge" concept enable more precise analysis and optimization of input data and processes.
To ensure the security and compliance of LLM applications, fortiss implements an advanced monitoring framework. This framework enables the enforcement of custom security and compliance policies, detection of hallucinations, privacy measures, and filtering mechanisms to avoid inappropriate content, ensuring the integrity and trustworthiness of AI applications.
In the field of industrial applications and edge computing, fortiss explores the use of LLMs to optimize automation and efficiency processes. The integration of LLMs into edge-cloud architectures enables the realization of energy-efficient processes, optimized production workflows, and intelligent automation across various industries.
fortiss investigates the application of LLMs to support neurodiverse individuals – particularly those with Asperger’s Syndrome or ADHD – in professional communication and social interaction. The focus is on developing solutions for communication facilitation, assistive applications, and AI-powered tools to promote accessibility and inclusion in the workplace.
Beyond general research on generative AI, fortiss develops specialized LLM-based solutions to provide small and medium-sized enterprises (SMEs) with a technologically sound and economically viable way to access Artificial Intelligence (AI). This is not only about providing practical AI solutions but also about their targeted adaptation to industry-specific requirements in order to systematically harness innovation potential.
A key research focus is the precise adaptation of LLMs for automating business processes, optimizing decision-making, and improving efficiency. Particularly in sectors such as manufacturing, logistics, and services, fortiss explores the use of LLMs to optimize supply chains, predictive maintenance strategies, and intelligent communication systems.
By developing flexible and scalable AI frameworks, fortiss lays the foundation for a sustainable and economically viable integration of LLMs into SMEs. This enables not only cost-effective implementation but also a long-term increase in competitiveness through adaptive and future-proof AI infrastructures.
LLM-based Process Optimization
The integration of Large Language Models (LLMs) enables the automation of repetitive business processes in SMEs, including content generation, handling customer inquiries, and document processing. fortiss explores the potential of this technology to optimize workflows and reduce operational costs.
Synthetic Test Data Generation through LLMs
LLMs play a crucial role in generating synthetic test data for AI validation processes, particularly in the creation of image and video data for testing perception systems. fortiss develops technologies that enable SMEs to independently and cost-effectively validate their AI applications, significantly lowering the entry barriers for AI adoption.
LLM Deployment on Edge Devices
For many SMEs, the implementation of LLM applications on resource-constrained devices is of central importance. fortiss is researching the optimization of LLMs for operation on devices with low computational power and reduced energy consumption to ensure resource-efficient and high-performance usage.
