Elexit-Builder2d

Home
Build
Challenges
Projects

Inspiration

A question before you start: Why don't all projects succeed the first time in engineering? Every engineer has countless failures before their project succeeds. The result? Well they finish their project, but many people quit after initial failure. Psychology proves that more than 90% (93% to be exact) people feel like they want to quit a task just because of initial failure.

It's not because the builder just messed up or he didn't follow the steps. It's because of the lack of experience, sufficient & personalized help people fail, and mainly because many people don't realize what they are actually doing; psychologically, after a set repeats the brain does a task autonomously (unconsciously for the human). They don't really know what they are doing and mainly they don't know that they need help. I want to eliminate that, just imagine people doing what they truly love instead of just leaving it just because of lack of help. I made this app so people, engineers in this case, can truly enjoy what they want.

What it does

Elexit is an AI-powered engineering workspace that combines learning, building, debugging, and experimentation into one platform.

This is main point of the idea of Elexit, one has to just type the materials the have or the project they want to do, the AI produces you a couple for projects for your chosen difficulty rate. After choosing a project the user wants to do, the AI produces build instructions with both text and image instructions so that the user can understand just by looking at the image, because scientifically humans tend to see images and understand more. The user can follow the steps and build their desired project. This helps reduce the failure of their efforts.

Objections

-One might ask "Why? Why do I care about people quitting after their failure? They need failure right?"

That is an interesting question. Say there are 100 people and their goal was to become an engineer, but because of failures, realistically, about 68% (68 people) quit. 68 people just gave up on their goal just because of failures, that's a lot. Just imagine you're one of the people who quit just because you didn't have some good help. As per the second question, it is not that I am just removing failure for people using this, I'm just postponing them for them to experience it in the future.

How I built it

This is actually like the 6th version of Elexit that I built from scratch (but it was coded after this hackathon started - the 6th version). I made this idea in March of 2026, but actually built it in end of April. From then, I kept on improving it until early June, then started again in mid July. The First version was called Elexit. Originally it was the same as Builder 2D; next was Elexit v2.1 with Systems and UX update; then, Elexit v3.1 with Improved layout and personalized bots; then, Elexit v3.8 with the builder at it's peak state; then, Elexit v4.1 (named Elexit - Midnight) with the addition of a newer model (local), features (digital twins), and utilities (labs). Finally, Elexit v5 with all the things that you see today.

As for the building part, I used Claude code, Codex, and other agentic AIs as well as myself doing a bit of the code. First I designed the web page in Canva to see it for myself. Then I implement the basic webpage code (in html, css, and JS). After that, I gather AI API keys to implement them in the project. Then, I leave the AIs to perfect it, rather I used them to accelerate development, while I designed the architecture, refined the interface, tested the application, and iterated on the user experience.

Challenges I ran into

I ran into a lot of challenges, the main ones being AI API credits limit, test failures, UI bugs, implementing the functionalities, and actually working.

-AI API credits limits: I knew that I was going to need a lot of credits, but in the project my GIT models stopped working. Then my Gemini models also stopped due a server issue from their side, it happened a couple of times. Then, I had to use a local LLM to run the application, but it fries my GPU really quick (it used to hit 80 C).

-Test Failures: Half of the times I test the application, it used to fail. Let it be server issues, model overloads, or payment credits, it used to fail because of them. But when it worked, it wouldn't be in the format of my speculation. Sometimes, it used to give some stupid results.

-UI Bugs: These were at the pinnacle of my challenges when I was trying to build Elexit. They would come out to my taste or they would just simply suck and don't please the user. I had to spend hours fix UI bugs, and there still are some UI bugs left.

Accomplishments that I'm proud of

First of all, I'm very proud to announce that my creation will be launched in the near future to help fellow engineers achieve what they seek. As for my accomplishments, I think just making a idea and building it is a real accomplishment, like many people make ideas, but less than half (about 32% to be exact) actually bring it to reality. I'm proud of doing that. Another accomplishment I'm proud of is I learned some new stuff (like fronted dev, servers, AI APIs, and a lot more).

What I learned

I've learned a lot. Like how to use AI APIs, how to implement them into the project, how to code them. I also learned that coming back to the project is a really hard task. I've also experienced the "failure" part for my self and I learned that with just a little bit of help I can do it. I learned that many people are out there without knowing what's limiting them, I wanted to help them.

Now speaking conceptually, I've learned some stuff. I learned prompting Agentic AIs, I have learned JS in the journey, and I learned how to present and market, just a little.

What's next for Elexit

Well, what's next for Elexit is a lot. For example an addition of couple more modules:

-Elexit - Mechanics: This is sort of a physics simulator, CAD, and advanced engineering application. You can make simulations either by experimentation [lab] or coding it in here. There will be other features coming in the future.

-Elexit - Study: Pretty self explanatory. This is just a study space for questions, courses, and experimentations.

-Elexit - Builder 3D: Currently Desktop exclusive for rendering 3D cad files. I might add it to web in the future with limit and low poly cad files for better performance. (extra info: For every project, the application pulls cad files from the cloud and attach it to the project, because storing all the models in it would use a lot of memory).

-Elexit - General: This is just a generalized form of all the general features of all the modules, but there are stuff exclusive to each module.

-Elexit - Desktop: This is my desktop app for all the modules in an app form, instead of a website. This is add more features.

Credits

Claude code & Codex : for coding the majority of the application.
Google studio, NVIDIA, GIT HUB, Groq, Ollama, and OpenRouter: for providing me with the AI APIs and local LLMs.
My Friend: for reviewing the project and giving me feedback.
Medium (link): for giving me scientific data.
Myself: for making this real.