Programmatic assessment for learning (PAL) involves programmatically structured collection of assessment data for the purpose of learning. In this guide, we examine and provide recommendations on several aspects: First, we review the evolution that has led to the development of programmatic assessment, providing clarification of some of its terminology. Second, we outline the learning processes that guide the design of PAL, including distributed learning, interleaving, overlearning, and test-enhanced learning. Third, we review the evolving nature of validity and provide insights into validity from a program perspective. Finally, we examine opportunities, challenges, and future directions of assessment in the context of artificial intelligence.