Machine Learning (ML) algorithms can enhance supply chain efficiencies in agribusiness. Despite their transformative potential, the adoption of ML in agribusiness is low. This paper is based on a systematic literature review (SLR), through the guidance of Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) model. Included in the SLR are 40 peer-reviewed studies published in IEEE Xplore, Springer Link, Science Direct, and ACM Digital between 2020 and 2024. Findings of this study reveal that Decision Trees and Neural Networks are used for yield prediction and market trend analysis, while Reinforcement Learning (RL) excels in dynamic logistics optimization, reducing transportation costs. Support Vector Machines (SVM) improve pest detection and soil quality assessments. Geographically, Asia dominates research output (55% of studies), with India and China focusing on crop yield optimization and resource allocation. However, technical challenges persist: Neural Networks face “black-box” interpretability issues, limiting stakeholder trust, while RL demands extensive training data and computational infrastructure. Data scarcity and siloed collaborations between agronomists and technologists hinder progress. These findings have implications on policymakers, by highlighting the need for standardized data-sharing frameworks and subsidies to incentivize AI adoption among small-scale farmers. The findings of this research inform researchers to prioritize research on hybrid models that balance accuracy with transparency, such as explainable AI (XAI) for crop health monitoring and low-cost algorithms like K-Means Clustering for demand segmentation. Stakeholders in the agribusiness can use these findings to make informed decisions regarding investment in low-cost emerging technologies for enhancing supply chain activities in agribusiness.

Keywords: Artificial Intelligence, Machine Learning, Agribusiness, Supply Chain Optimization, Sustainability, PRISMA.