Automating Form Completion with Large Language Models

Manual form filling, especially when technical data must be retrieved from various documents, remains tedious and error-prone. We present a prototype system using Large Language Models (LLMs) to automate and streamline form completion, demonstrated through a case study in the MICS project, which supports circular economy practices. Our approach combines automatic extraction with human-in-the-loop review, significantly reducing user effort and time while maintaining accuracy. User study results indicate improved usability and satisfaction, but highlight the need for safeguards against AI “hallucinations” and further integration.