TL;DR
AMAI starts the GenKiwi research project as a technology partner and coordinator. Together with Karlsruhe University of Applied Sciences and industry partners, we are developing an AI system that not only searches documents, but also actively secures and makes use of implicit experience from employees. The project is funded by the BMBF and ESF Plus.
New ways of managing knowledge
When long-time employees leave a company, more often goes than just one worker: valuable experience built up over years disappears. For SMEs in particular, this “knowledge drain” in the context of demographic change is one of the biggest challenges of the coming years.
There are many documentation systems, but they often fail due to the time factor or the complexity of properly storing unstructured knowledge (emails, notes, thoughts).
This is exactly where the joint project comes in GenKiwi (Human-centered generative AI for quality-assured retrieval of in-house knowledge). AMAI acts as a technology partner and project coordinator to develop new ways of knowledge management together with partners from research and industry.
More than just a search: An active knowledge agent
GenKiwi's goal is ambitious: We are developing an AI-based knowledge management system that goes far beyond classic enterprise search. Instead of just responding to inquiries, the system acts proactively.
What makes GenKiwi special:
- Capturing implicit knowledge
The biggest capital of many companies is not in any handbook. It is the “gut feeling” of an experienced master or the trick that an engineer applies to specific problems. Our agent system is trained to identify gaps in knowledge. It actively approaches employees and asks specific questions in order to secure this undocumented experience before it is lost. - Human-in-the-loop & learning processes
AI does not simply serve as an information engine that does the thinking. In the sense of a human-centered approach, the system promotes reflection through inquiries. Employees are at the center: AI supports learning processes and helps to better understand relationships. - Intelligent, natural interaction
Thanks to large language models (LLMs), employees with company knowledge can speak as naturally as with a colleague. The system makes unstructured knowledge available across departmental boundaries without having to know which folder a file is in.
Strong partners for SMEs
Such a project requires diverse perspectives. We are proud to be part of a strong consortium that combines research and industrial practice:
- AMAI GmbH: As project coordinator and technology partner, we bring our expertise to Generative AI, Retrieval Augmented Generation (RAG) systems and software engineering. We bridge the gap between current research and productive application.
- Institute for Learning and Innovation in Networks (ILIN) — Karlsruhe University: Our research partners contribute deep expertise in human-centered work design and competence development.
- Houses + Renner KG (application partner): At the CNC technology specialist, we optimize the calculation of offers. The AI analyses historical project data and technical drawings to estimate processing times more precisely.
- ANSMANN AG (application partner): For the mobile energy expert, the focus is on a central knowledge platform that bundles decentralized knowledge and makes it easy to find it efficiently.
Funded by BMBF and ESF Plus
The GenKiwi project is part of the program “The future of work” through the Federal Ministry of Education and Research (BMBF) and through the European Social Fund Plus (ESF Plus) promoted. We would also like to thank the Project partner Karlsruhe (PTKA) for a constructive start.
Outlook
The starting signal has been fired. In the coming months, we will be working on developing AI agents and testing them directly in our partners' application scenarios.
On the project website GenKiwi — Human-centered generative AI for knowledge management receive regular updates on technical milestones and practical findings.
