AI is becoming one of the most important technologies students can learn, but the way we teach it has barely changed.
Most AI education still follows the same pattern: watch a lesson, read some documentation, follow a tutorial, and answer a few questions.
That can teach you what AI is.
It doesn't necessarily teach you how to use it.
A student can understand what a context window is, know what an AI agent is, and explain how image generation works, while still struggling to actually build something with those concepts.
That gap is what led me to build VersedAI.
The Problem
Learning AI is fundamentally different from learning a static subject.
The field changes quickly. New models, tools, workflows, and techniques appear constantly. Because of that, memorizing definitions is not enough.
Students need to develop the ability to experiment.
They need to be able to try something, see what happens, make mistakes, understand why it failed, and try again.
Traditional educational content isn't particularly good at creating that environment.
A video can explain prompting.
A student actually learns prompting when they write a prompt, get a bad result, figure out why it was bad, change it, and try again.
That is the learning experience I wanted to build.
Learning by Doing
VersedAI is a hands-on AI learning lab designed for high-school students.
The basic loop is simple:
Learn → Build → Get stuck → Get a hint → Try again
Instead of separating theory from practice, VersedAI connects them.
A student learns a concept and immediately gets an opportunity to apply it.
If they get stuck, they can ask an AI tutor for help.
The important part is that the tutor isn't supposed to simply solve the exercise.
It provides guidance and hints so the student still has to think through the problem.
The goal isn't to make the exercise easier.
The goal is to make the student better.
Four Ways to Explore AI
VersedAI currently organizes learning into four areas.
AI Fundamentals
This introduces the basic concepts behind modern AI and gives students a foundation before moving into more practical topics.
Image Studio
Students experiment with image generation and learn how different instructions affect the output.
Instead of simply generating an image, the experience focuses on understanding how to communicate effectively with generative models.
Context Lab
Context is one of the most important concepts in modern AI applications.
Students experiment with how information is provided to models and how changing the available context can change the result.
AI Agents
Students move beyond simple chatbot interactions and start exploring agents that can reason through tasks and use tools.
This is where AI becomes less about asking a model a question and more about designing systems that can accomplish a goal.
The AI Tutor
One of the most important design decisions in VersedAI was how the AI tutor should behave.
It would be easy to build a tutor that simply gives students the correct answer.
That would also defeat much of the purpose.
If a student asks for help and immediately receives the solution, they may complete the exercise without actually understanding it.
So the tutor is designed around guided problem solving.
When a student gets stuck, the system can provide progressively useful guidance rather than immediately revealing the solution.
The student remains responsible for completing the task.
That creates a different relationship with AI.
Instead of using AI as a shortcut, the student uses AI as a teacher.
The Technology
VersedAI is built around Google's AI ecosystem.
The application uses Gemini for the core model interactions, Google ADK for agent-based functionality, Gemma for additional model capabilities, and Imagen for image generation.
The application itself uses Next.js on the frontend and FastAPI on the backend.
The infrastructure runs through Vertex AI and Cloud Run.
The architecture is intentionally split between the learning experience and the AI services.
The frontend handles the interactive educational experience.
The backend manages the application logic and communication with the AI systems.
This makes it possible to evolve the educational experience without having the entire application tightly coupled to a single model interaction.
Why Build This?
The goal isn't to create another AI course.
There are already thousands of AI courses, tutorials, videos, and documentation pages.
The harder problem is creating an environment where someone can actually develop intuition for working with AI.
I think that intuition comes from experimentation.
You need to see what happens when you change the prompt.
You need to understand why an agent failed.
You need to discover that giving a model more context doesn't always make the result better.
You need to build something that doesn't work and then figure out why.
That's difficult to teach through passive content.
It's much easier to learn when the system itself gives you something to experiment with.
What I Learned Building It
The biggest lesson from building VersedAI is that an AI educational product shouldn't just put an LLM in front of a student and call it a tutor.
The interaction design matters.
The model needs to know when to explain, when to ask a question, when to provide a hint, and when to let the student struggle.
That last part is especially important.
Good education isn't about removing every obstacle.
Sometimes the obstacle is the learning.
VersedAI is an attempt to build an environment where AI helps students overcome the right obstacles without removing the thinking entirely.
What's Next
There is still a lot to improve.
I'd like to make the learning paths more adaptive, improve the tutoring system, add more challenging projects, and eventually make it possible for students to build more complex AI applications as they progress.
The long-term idea is simple:
AI shouldn't just teach students about AI. It should help them learn how to build with it.
That's what I'm exploring with VersedAI.
Try it here:
And the project is open source:
Top comments (0)