Case study AI-deation
To make the identification of innovative research ideas more efficient, fortiss developed the “AI-deation” platform for Siemens. The system uses language models to synthesise scattered knowledge, recombine existing expertise and generate entirely new research approaches. Through the use of data protection-compliant Small Language Models (SLMs), a scalable and secure innovation environment is created that helps experts move more quickly from a technological capability to a concrete, effective use case.
Challenge
In a rapidly changing technological landscape, it is becoming increasingly difficult to identify highly relevant research opportunities and use cases in a timely manner. Traditional ideation processes are often time-consuming, heavily dependent on the availability of individual experts, and limited by human capacity to process vast amounts of interdisciplinary knowledge. Siemens faced the challenge of scaling innovation processes across different business units without compromising on data protection or cost efficiency.
Solution
The solution, an intelligent ideation platform, is based on a hybrid architecture comprising powerful Large Language Models (LLMs) and specialised, privacy-friendly Small Language Models (SLMs).
The system acts as a digital catalyst for human creativity: it integrates seamlessly into existing processes and helps researchers to recombine existing knowledge or explore entirely new ideas beyond known publications. In this context, fortiss contributed in particular its expertise in optimising the models and ensuring data integrity.
Result
- Provision of a platform that generates use cases and research ideas based on existing Siemens expertise
- Systematic integration of technological knowledge from different specialist disciplines to tap into new market segments
- Increased efficiency in meetings: Teams can work directly with an AI-generated draft (‘scaffold’) instead of spending valuable time summarising what is already known
- Creativity boost: Generating novel approaches that go beyond the current state of the published literature
- Data protection-compliant innovation: Use of SLMs to process sensitive corporate data within a secure infrastructure.
Outcome
The AI-deation project demonstrates that generative AI does not replace the human ideation process, but rather significantly accelerates it and enhances its quality. By combining LLMs and SLMs, interdisciplinary synergies can be efficiently harnessed in large corporations such as Siemens. fortiss’s expertise has made it possible to integrate complex language models into an industrial context in such a way that they serve as a reliable tool for the research of tomorrow.
Project partner
