Document AI or Document Intelligence has emerged as a critical field that seeks to automate the analysis of documents using advanced natural language processing and computer vision techniques. Key tasks in this domain include understanding forms, receipts, and invoices, analyzing document layouts, extracting visual information, answering questions based on document visuals, and classifying document images, among other activities. This process is commonly known as Visual Document Understanding (VDU). Its goal is to examine scanned or digital business documents, automatically extracting and organizing structured information for various business applications. VDU is challenging because these tasks involve interpreting the text content, vision, and spatial relationships of the document. They may focus on images, identifying and labeling areas like fields, headers, and paragraphs, or they may focus on text, classifying individual words and phrases. This paper offers an overview of existing research in the field of Visual Document Understanding, providing a brief review of sophisticated models, their diverse architectures, associated tasks, and benchmark datasets. We explore recent advancements and breakthroughs in VDU and finally, the potential future directions for research in the field of visual document understanding.