The physicochemical properties of natural chemical compounds serve as a foundation for the development of novel drugs and innovative therapies. While several databases describe properties of natural products, their applicability and data accessibility are limited. Thus, the lack of accessible data represents a significant challenge in developing drugs based on natural compounds. Although chemical properties of natural compounds can be determined experimentally, this approach requires costly materials and procedures. In silico alternatives for drug analysis and pharmaceutical design cycles represent an interesting, simpler, and less expensive option for natural compound-based drug discovery. This article examines in silico methods for the characterization, design, and optimization of natural compound-based drugs derived from food. The review focuses on how in silico-based tools, such as machine learning, computer-based mathematical modeling, homology prediction, docking, molecular dynamics, and simulated molecular evolution events, are used to optimize natural compound testing and design. The in silico bioactivity predicted properties for peptides and secondary metabolites are discussed. In silico analysis is also explored as a tool to predict the antioxidant, antidiabetic, antimicrobial, and cardiovascular effects of natural compounds from foods. The approaches here presented can help speed up the discovery and development of natural compound-based drugs for therapeutic use.