Timeframe and Context

The project started in March 2026 at the Institute for Applied AI and Robotics.

The goal was to develop an AI-based assistant system that makes domain-specific mechanical engineering knowledge accessible and supports technical questions during the development process. In the first step, the focus was on integrating relevant standards and technical information into the chatbot so that they can be queried in context and used for development tasks.

The chatbot was designed as a prototypical solution to investigate how Large Language Models can be used to support mechanical engineering development processes.

As the sole developer, I bore full responsibility for the project, from conceptual design to the development and evaluation of the of the assistant.

Implementation and Tech Stack

For the implementation, a RAG-based chatbot application was developed in Python. Using LangChain, documents from the field of mechanical engineering were processed, indexed, and made available for context-based responses to user queries.

The system was based on a locally operated Large Language Model connected via Ollama. This allowed the prototype to be operated in a privacy-friendly way and made it particularly suitable for handling sensitive or internal technical information.

Challenges and Results

A central challenge was preparing complex technical documents and standards content in such a way that they could be reliably used to answer domain-specific questions. It was particularly important that the chatbot did not only generate general answers, but referred to the integrated documents in a context-aware manner.

By using RAG, a first prototype was developed that makes mechanical engineering domain knowledge accessible in a dialogue-based system. The solution demonstrates how LLM-based assistant systems can support technical development processes by making relevant information easier to find and simplifying work with extensive technical and standards documents.

The project therefore forms a foundation for further applications, such as supporting design engineering, technical documentation, standards research, or knowledge-based assistant systems in mechanical engineering.