Abstract:

Digital work instructions are widely used on production floors, yet often assume native-language proficiency and reading fluency that do not align with training-oriented industrial workplaces. This creates barriers for workers who are simultaneously developing professional and language skills. We present a solution that augments existing digital work instructions with an AI-driven support layer for in-situ learning and personalization while minimizing disruption to production workflows. The approach preserves original documents and overlays optional assistance, including simplified text, contextual vocabulary, and automatically generated quizzes. Content is pre-processed through OCR, language-model-based cleanup, and task-specific enrichment, enabling lightweight run-time interaction within the original instruction interface. Developed in close collaboration with a manufacturing company, the system is presented as an engineering case study of AI-supported augmentation in real-world industrial contexts. We contribute (1) an integration pattern for embedding AI-supported learning into existing industrial documentation, and (2) insights into balancing assistance with workflow continuity.

Cite (BibTeX):

@inproceedings{hendrikx_eiseait2026,
  author    = {Hendrikx, Maria and Geurts, Eva and Luyten, Kris},
  title     = {Words at work: In-context Dutch learning through low-interruption AI integration in digital work instructions},
  booktitle = {The fourth workshop on engineering interactive systems embedding AI technologies (EISEAIT)},
  year      = {2026}
}