Collaboration

Here is a blog post by edutool member Ahmet Onur Agca, sharing his own experiences with vibe coding for education:

For a long time, many of us have wanted to bring more interactive elements into our teaching, but we have faced significant hurdles along the way. We often find that gamification usually fails in traditional settings because our curriculum evolves weekly, whilst traditional development takes months. Frequently, the resulting educational games sometimes felt like expensive spreadsheets with a skin, bogged down by over-engineered user interfaces that distract from our true pedagogical goals. Furthermore, the high costs and technical debt required for these disposable learning modules can be overwhelming for any department.

However, I want to share a wonderful shift I have been exploring in my own practice: the era of vibe-coding.

Fullstack Educator

We no longer need to be software developers to create engaging, dynamic tools. Through vibe-coding, we can build with words, turning plain language into working apps. It is a highly beginner-friendly process where the AI handles the heavy lifting, allowing us to focus on guiding the vision. This shift empowers us to become Fullstack Educators, gracefully managing both the frontend visuals and the backend logic of our learning materials. We can move swiftly from an initial idea to launching a minimum viable product or a full workflow, enjoying great flexibility across web and mobile platforms.

Beyond simply making learning more engaging, the real value of these tools lies in the research and learning analytics. When we design these interactive scenarios, every button click becomes a granular data point reflecting student cognition, thinking, learning, and decision-making. This allows us to conduct A/B testing in real-time, instantly swapping theoretical strategies or pedagogical methods across different student groups to see what resonates best. Additionally, we can turn session logs into automated, personalised feedback loops. Ultimately, this provides us with observable and analysable granular data to truly understand our students' learning journeys.

A Gentle Roadmap to Getting Started

If you are curious about trying this in your own modules, here is a simple roadmap based on my experience:

• Ideation: The most important step is building the pedagogy, not the page. Design your independent variables (student choices) and dependent variables (consequences). Try sketching out actions, reactions, and success states using a simple mind map or PowerPoint. For example, in a math game, instead of asking students to "solve for x," design a scenario where they must "apply x to survive the scenario".
• Build: Think about your game design, including objectives, randomisation, and cohort size. You can utilise backend databases like Supabase, Firebase, or Google Cloud. For the frontend, you can create a draft sketch via GenAI or describe what is in your mind to a UI designer tool like Vercel.
• Vibe-Code & Launch: Translate your pedagogical "vibe" into structural components using interfaces like Cursor, VS Code or Antigravity. I frequently use Claude Code as the AI language. Once you create a draft of your game, you can transfer your local files to a GitHub account, host them on platforms like Vercel, and then even purchase your own domain.
• Test & Refine: Always test the MVP yourself first, and then share it gently with your community for feedback.

CAUTION: Navigate the Landscape Carefully!

Of course, this new frontier comes with its own set of challenges that we must navigate thoughtfully.

• Monitoring Costs: API token usage is a Black Box that needs active monitoring. I highly recommend using free versions of tools like Supabase and Vercel until you are completely sure about the safety and conscious maintenance of your database and API keys.
• Ensuring Accuracy: We must always implement verification layers for factual accuracy, as AI drift can sometimes corrupt the logic.
• Staying Flexible: Because the structural evolution of these tools is happening every week, it is best not to get stuck with annual memberships as long as you are not getting for free. Do not invest too much time and effort into becoming an absolute expert on any single platform, as all becoming easier to adapt, learn and apply in time.
• Considering Ownership: Finally, there is an ongoing and important conversation about who actually owns the code when the ‘developer’ is simply a prompt.

The end of the developer bottleneck is opening up wonderful possibilities for our pedagogy. I encourage you to start small, perhaps by mapping out a single scenario for your next lecture. Let us explore this exciting space together.