Learn By Building

Inspiration:

Most programming education is passive. Learners watch videos, copy tutorials, and complete guided exercises, yet still freeze when asked to build something from scratch.

We wanted to close that gap.

The inspiration behind Learn By Building was simple: the best way to learn programming is to actually build things. Instead of following a fixed course from chapter 1 to chapter 10, learners should practice applying their knowledge, make mistakes, receive useful feedback, and gradually take on harder challenges.

We imagined an experience that feels less like taking a course and more like having an AI mentor who continuously gives you the next thing to build.

What it does:

Learn By Building is an AI-powered, project-based learning experience for programming.

A learner chooses a technology they want to learn, such as React, and starts with a small project appropriate for their skill level. They build the project themselves, submit their code, and receive concise, mentor-style feedback.

The AI focuses on helping learners understand:

What they did well What is wrong or could be improved Why the issue matters How they can solve it themselves

Rather than immediately giving away the answer, learners can use hints when they get stuck. As they demonstrate their understanding, they unlock increasingly difficult projects.

The core loop is simple:

Build → Get Feedback → Improve → Unlock → Build Again

How we built it:

We designed the experience around a minimal learning loop rather than a traditional course structure.

The interface is intentionally clean and focused on the current project. Learners receive a clear goal, requirements, and optional hints before writing their own code.

We use AI as a mentor throughout the process. It evaluates submitted work, explains mistakes in context, and provides actionable feedback without taking the learning process away from the learner.

Progression is also tied to demonstrated ability. Instead of simply completing lessons in order, learners move forward by successfully applying what they've learned in real projects.

Challenges we ran into:

One of the biggest challenges was finding the right balance between helping learners and letting them struggle productively.

An AI can easily solve a problem for someone, but that doesn't necessarily mean they've learned anything. We wanted feedback to guide learners toward the solution rather than simply replacing their work.

Another challenge was designing projects that are small enough to feel approachable while still being meaningful enough to teach real skills. Each challenge needs to introduce something new without overwhelming the learner.

We also had to think carefully about progression. If projects become difficult too quickly, learners get stuck. If they remain too easy, they stop learning. The goal is to continuously provide a challenge that is just beyond the learner's current ability.

Accomplishments that we're proud of:

We're proud of creating a learning experience that puts building at the center of programming education.

Instead of asking learners to consume more content, Learn By Building encourages them to actively practice the skills they want to develop.

We're also proud of keeping the product focused. The experience is built around a small number of core interactions—projects, hints, code submission, feedback, and progression—without adding unnecessary complexity.

Most importantly, we created a foundation for learning that feels closer to having a personal programming mentor than taking another online course.

What we learned:

We learned that learning to code isn't just about understanding concepts. It's about developing the confidence to apply those concepts independently.

We also learned that AI is most valuable in education when it doesn't simply provide the answer. The better experience is one that asks learners to think, gives them the right amount of guidance, and explains why something works or doesn't work.

The most important lesson was that progression should be based on what someone can actually do—not how many lessons they've clicked through.

What's next for Learn By Building:

Our next step is to expand the project progression and make the AI mentor better at understanding each learner's strengths and weaknesses.

We want Learn By Building to eventually support more technologies and provide a personalized path where every project builds naturally on the skills learned in previous ones.

The long-term vision is simple: replace passive programming education with a continuous cycle of building, feedback, and growth.

Built With

• auth.js
• css
• next.js
• postgresql
• prisma
• provider-agnostic
• react
• tailwind
• typescript
• vercel
• zod