Personalized learning and assessment tools have emerged as an approach to meet the needs of (neuro)diverse learners at scale. To ensure effective learning opportunities, it is crucial that educational tools include a balanced interplay between machine-driven adaptation (i.e., adaptivity) and learner-driven adaptation (i.e., adaptability). We first briefly review the existing principles and guidelines for universal learning and assessment tools that involve both adaptivity and adaptability. Subsequently, building on existing principles and guidelines, we propose a framework for designing evidence-based dynamic personalization for learning and assessment tools (ED-PLAT) that optimize adaptivity and adaptability to support (neuro)diverse learners’ motivational, affective, and cognitive processes. As a case study, we then leverage a personalized learning and assessment tool aimed at optimizing the learning of factual knowledge, highlighting how its features can be developed and modified to support the motivational, affective, and cognitive processes of diverse learners, specifically learners with dyslexia and attention-deficit hyperactivity disorder (ADD/ADHD), and outline research directions for evidence-based design decisions. Overall, our contribution comprises a review of existing guidelines and identification of steps forward, attending to universal design principles that maintain learner agency while implementing machine-driven adaptation to ensure effective and efficient learning and assessment for all learners.