This challenge turns vibe coding toward the practical: building a small tool that handles a repetitive task in your teaching life. Along the way, you’ll confront one of the central tensions of AI-assisted creation the gap between it works and I understand it.

Vibe coding shines at practical problem-solving. With a few prompts, you can build a tool that automates something tedious no setup, no syntax, no Stack Overflow. For educators drowning in small repetitive tasks, this is genuinely appealing.

There’s a catch worth feeling firsthand. When AI writes the code, you can end up with a working tool whose inner logic you can’t fully read. This is the heart of a real debate about vibe coding: the gap between a thing that runs and a thing you actually understand. Before we ask students to navigate that gap responsibly, it’s worth experiencing it ourselves.

Challenge

Build a small tool that streamlines something you do often, without writing the code yourself. Then interrogate it: not just “does it work?” but “do I understand it well enough to trust it?”

Pick your scenario

Pick a real, recurring annoyance—the smaller and more specific, the better.

• A calculator or converter (a weighted-grade calculator, a reading-time estimator)
• A generator (flashcards from a list, discussion questions from a topic, randomized groupings)
• A simple dashboard that displays or sorts information you track regularly
• An organizer (a checklist, a cold-call name-picker, a class timer with stages)

Instructions

1. Name a real annoyance: Choose a task you genuinely repeat, so the stakes feel real.
2. Choose your pathway: Stay no-code/low-code with a tool that previews apps from chat (like Claude), or if you have some experienceask the AI for the actual code and run it yourself. Tools ChatGPT, Claude, Co-pilot
3. Prompt for a working draft: Build me a simple [tool] that does [X]. I’m not a programmer make it run in the browser with no setup, and explain in plain language what it does.
4. Iterate by describing the problem, not the fix: Tell the AI what’s going wrong (“it breaks when the list is empty”) rather than how to solve it, and see whether it can diagnose its own work.
5. Try to read the logic: Ask the AI to walk you through how the tool works, step by step. Notice how much you can follow and where you simply have to take its word.
6. Stress-test it: Feed it strange or empty input, edge cases, or something you’d never expect a student to type. Watch what breaks.

Reflect on trust and understanding

• Does the tool actually work? How do you know did you verify it, or just trust that it ran?
• Could you explain to a colleague how it works? Where would you get stuck?
• Where would you not be comfortable relying on AI-generated code and why?
• What would a student need to learn to use tools like this responsibly rather than blindly?