The article presents a microservice architecture designed to enhance the operational efficiency of document AI systems, focusing on OCR and large language models. It highlights key design decisions and findings that can help practitioners effectively deploy these technologies at scale.