Document administration at educational institutions, such as Esa Unggul University, often faces challenges in filling out forms manually, making them time-consuming and prone to human error. This research develops a form filling automation system based on Optical Character Recognition (OCR) and Large Language Model (LLM) technology. OCR technology is used to extract text from certificate documents automatically, while LLMs are used to understand context and identify specific information, such as names, activity titles, and execution dates. The research method used a waterfall approach, with Fishbone Diagram analysis to identify the root of the problem and system design using General OCR Theory (GOT) for text extraction as well as the Llama 3 model for data interpretation. The system is tested with a variety of certificate formats to evaluate the level of accuracy and efficiency compared to manual methods. The results showed that the system was able to achieve an accuracy of more than 90%, reduce input errors, and speed up the form filling process. He concluded that the developed system has succeeded in improving the efficiency, accuracy, and reliability of the administrative process at Esa Unggul University. This system is expected to be implemented to support more efficient and accurate campus administration, as well as be the basis for the development of document automation technology in the future.