Brick Set
Brick Set is an interactive 3D film set built from Lego-style bricks. Students walk around the set, meet 25 crew positions from director to intimacy coordinator, and hear each person explain, in first person, how they behave on set and who they answer to. A second mode, "Who do I ask?", gives students a real on-set situation (a shadow is falling across the lens, the camera card is full) and asks them to walk to the right person and click on them. Each round draws 10 situations from a bank of 31.
I built it in conversation with Claude, Anthropic's AI assistant. I started with a plain-language brief: a walkable Lego-style set where each crew position explains its etiquette. Claude wrote the code (HTML, CSS, and JavaScript, using the open-source three.js library for the 3D), and I did the director's job: set the vision, review each version, and say what wasn't working. The first version was a flat 2D map. I told Claude that some elements overlapped and were hard to read, and it rebuilt the set as an orthographic 3D scene with a rotating isometric camera, spread the departments onto separate baseplates, and added name tags that move out of each other's way. Then I asked for an educational layer, and we added the "who do you go to" challenge, with hints written for the mix-ups students actually make. The finished build is a single file that drops into a website or LMS page.
Brick Set is a Lego-style 3D film set where students meet 25 crew positions, learn each one's set etiquette, then play "Who do I ask?" by walking to the right person for real on-set situations. Built with Claude and directed by an instructor, it runs in any browser and drops into a website or LMS.
How it was built
How I used AI
AI worked as a build partner and rapid prototyper. I described problems in plain language, and it handled the implementation and tested each revision. That took the project from idea to working learning tool in a few sessions, not months. The expertise stayed with me: what is accurate, how crews really work, and where students get confused. Because the AI drafts the crew rules and quiz scenarios, I review them against professional practice, as I would any draft, before students use them.
Learning by doing. Students practice the chain of command spatially instead of memorizing a list of titles.
Low-stakes practice with feedback. Each question gives two guesses, and a wrong answer explains why that person is not the right one. For example, the DP designs the lighting, but the gaffer's crew carries it out.
Ready for the real set. Students learn the vocabulary, the etiquette, and who to ask first before their first lab shoot.
Flexible. Use it as a pre-lab warm-up, a review, or a self-paced intro. It runs in a browser on laptops, tablets, and phones, with no install and no accounts, and the app itself stores nothing about students.
A model for what's possible. With clear learning goals and an AI partner, an instructor can build a custom teaching tool without a developer.