Fraunhofer Institute for Manufacturing Engineering and Automation IPA
GenAI - implementation of a knowledge database (RAG)
The results of generative AI (GenAI) cannot be evaluated by the models themselves. It is therefore necessary to develop clear evaluation criteria. The integration of different data sources and AI models in mechanical engineering is also a complex task. Our offering includes the implementation of a RAG system, which uses the technology behind chatbots to efficiently process internal data, such as manuals, error reports and service documentation, and to make this accessible. By using fine tuning, the models can be adapted to the specific vocabulary and content of your company.
Value proposition
A RAG system enables internal company data to be integrated and used efficiently, which helps to empower employees quickly and to increase productivity. With fine tuning, the models can learn specific vocabulary and content, thereby improving the accuracy and relevance of the results. This is particularly important in mechanical engineering, where the shortage of skilled workers and the quick onboarding process for new employees represent challenges.
Investment and scope of services
This project starts from 30,600 euros (for a simple use case) and comprises the following services:
- Defining challenges and targets: Creating a solid database.
- Examining appropriate AI models: Evaluation of a suitable model.
- Definition of the overall system: Determining the necessary system. Checking whether third-party solutions (such as Nuclia or open-source tools) can be used.
- Benchmarking: Evaluation of the performance of the new system.
Optional additional services
Integration into the overall system: Adaptation and implementation with a commercial solution.