You are currently viewing No Programming Experience Necessary: How I Build Interactive Learning Tools with Generative AI

No Programming Experience Necessary: How I Build Interactive Learning Tools with Generative AI

Like many faculty, I’ve long assumed that building custom digital tools was work that belonged to someone else. I am not a programmer, but over the past year, I developed a series of interactive web applications for my anatomy and physiology courses using generative AI.AI allows me to be the subject matter expert while it translates my instructional ideas into functional, web-based applications. This process has not only expanded the resources I can offer my students but also changed the way I think about what is possible in my courses.

What the process looks like

One of the first tools I built focused on the physiological mechanism involved in tropic hormone pathways. This is a particularly difficult concept for students as these pathways consist of a chain of hormone signals between the hypothalamus, pituitary gland, and target organs.

To build this tool, I simply logged into Claude and described in plain language how I wanted the tool to function. I also uploaded my own course materials so the content and vocabulary would align with what students were learning in class. I asked Claude to help me create an interactive experience where undergraduate students could move through the pathway step by step, predict what happens next, and receive feedback that explains not just whether they were correct, but why.

My initial prompt instructed Claude to provide me with an “interactive learning tool that would help my undergraduate anatomy and physiology students understand the hypothalamic-pituitary-thyroid axis.” I explained that I wanted them “to put the steps of the pathway in the correct order and understand the relationship between the components.”

Claude generated the initial code within a matter of minutes and provided a preview of the application. The initial output was a good starting point, but I wanted to add more complexity. Once the students understood the pathway itself, I wanted them to apply that knowledge to pathological scenarios, so I revised the prompt to include additional applications. I spent the next few hours testing the tool and flagging what was not working instructionally.

The process was iterative, requiring me to refine my prompts and test the revised output repeatedly. By the end of the day, I had a functional tool I could put in front of students, and at no point did I write a single line of code. When I felt confident in what I had created, I asked Claude to produce the tool as a SCORM package, which follows a standardized set of e-learning specifications that allow interactive content to be delivered through learning management systems. This allowed me to upload the tool directly to Canvas for my students to access like any other resource.

Getting it right

The first version of any tool that I create with Claude is never perfect. For example, there were several issues with the early version of my cardiac cycle tool. I first noticed that Claude had mixed up some of the answer key responses in the Wigger’s diagram section, associating correct physiological explanations with the wrong phases of the cycle. I caught these errors during my content review and described the problem directly, specifying which answers were incorrect and what the correct associations should be. Claude fixed them immediately. It should be noted that content accuracy demands careful attention throughout the development of these tools with generative AI. Platforms like Claude often produce plausible-looking content that is sometimes subtly wrong, so I thoroughly verify each tool I build before it reaches students.

I also encountered technical issues in the cardiac cycle tool. I noticed that as students scrolled through the question sets, the diagram scrolled out of view, making it difficult to answer questions about it. I simply asked Claude to make sure the diagram was visible at the top of the page while the students worked through the questions below it, and it fixed the problem.

To apply an additional layer of review, I asked my lecture assistants to test each tool before releasing it in the course. During the review process, it is important not to assume that the tool will work for everyone just because it worked during your initial review. Having people who were not involved in the design process test out the tools is imperative as they can uncover usability issues and problems caused by unexpected user interactions. For example, my lecture assistants found that if they placed a component incorrectly in the tropic hormone pathway tool, there was no way to remove it without restarting the entire pathway. I relayed that feedback to Claude, and it added a double-click function that returns a misplaced component to the panel. This adjustment allowed students to correct their work without starting the pathway over.

Student response

The student response was immediate. As soon as the tropic hormone pathway tool went live, students started asking when the next one would be available. Several told me that it was the first time these concepts had made sense to them—so I kept going. I moved on to other challenging concepts, building additional tools that allowed students to work through complex physiological processes sequentially.

Rather than asking students to memorize diagrams and mechanisms, I was able to offer them a tool to work through the concepts interactively, predicting what is happening at each phase before the answer is revealed. This allowed me to create the kind of structured, active practice that is nearly impossible to replicate in a lecture hall and that previously would have required a development team to produce.

One unexpected benefit has been the speed at which I am able to revise and adapt the tool to students’ needs. If a student flags something confusing, or I recognize that a concept needs a different entry point, I can describe the change I want and have a revised version within the hour. That kind of responsiveness has shifted how I think about course materials: less as fixed products and more as living resources that can evolve alongside student understanding.


Amanda Gunlefinger is a teaching assistant professor in the Department of Biological and Chemical Sciences at Thomas Jefferson University. Contact: amanda.gunlefinger@jefferson.edu.

The post No Programming Experience Necessary: How I Build Interactive Learning Tools with Generative AI appeared first on The Teaching Professor.