To develop expertise, novice chemistry students require deliberate practice with immediate and thoughtful feedback. Whereas opportunities for practice are often abundant, opportunities for feedback tend to be much scarcer. To address this challenge, we collaborated with students to develop a flexible and scalable online platform called "Alchemy." Students interact with the software through a series of prompts and multiple-choice questions, receiving targeted feedback at each step. Each student can navigate a scenario differently, influenced by their prior knowledge and misconceptions. In design, scenarios are instructor-created graphs that model their expert-level thought process­(es) and teaching experience. This graph format is a distinct and advantageous feature of Alchemy that supports linear, nonlinear, and cyclical progression through a concept or problem. Alchemy-based homework has been assigned in lecture and lab courses, at the sophomore and junior levels, with enrollments ranging from many tens to many hundreds of students. Approximately 80% of students rated Alchemy as very/extremely helpful to their learning, emphasizing the value of the step-by-step process and feedback. Alchemy is discussed in the contexts of other software platforms, its advantages and limitations, lessons learned in its initial deployments, and future directions for its development and usage.