Know exactly where you stand.
Know exactly what to study next.
Add your subjects and topics so Study Triage can calculate your academic risk.
Click "Add Subject" to start building your semester profile.
Add at least 2 subjects to unlock the dashboard.
Subjects ranked by academic risk — highest priority first
How I used AI, what it suggested, what I changed, and what I decided on my own.
I built Study Triage for the BET Software CRASH 2026 competition. The brief required using AI as an assistant — not as the main builder. That is exactly how I treated it. I used AI to explore ideas and challenge my thinking, but every final decision about what to build, how it works, and why it looks the way it does was mine.
This report is honest. It shows the prompts I actually used, the suggestions I rejected, and the places where I disagreed with AI and went my own way.
I started with real problems, not a solution. I used AI to brainstorm what first-year students actually struggle with across time management, academic performance, and mental load.
From the list AI gave me, I picked the problems that felt most genuine and that I could actually solve in a single HTML file:
AI also suggested group coordination and peer accountability features. Rejected — no network access in a single HTML file, so those would have been fake features.
I asked AI to suggest application categories I could build. It gave me four directions:
| AI Suggestion | My Decision | Reason |
|---|---|---|
| Priority engine (ranked task list) | Modified | Used the scoring logic but rejected the plain list output — would not stand out visually or score well on GUI marks |
| Exam week scheduler | Rejected | Scheduling without real calendar access felt shallow |
| Sorting algorithm visualiser | Modified | Kept the sorting idea but embedded it inside the app logic instead of making it the main feature |
| Subject health dashboard | Accepted | Used as the visual shell — the dashboard format communicates risk clearly at a glance |
My final concept combined the dashboard with the urgency score engine as the brain behind it. I also independently decided to structure the app into three connected views — setup, dashboard, deep dive — so the sorting algorithm is visible at multiple levels of detail, not just hidden in the background.
This is the core intellectual contribution of the app. AI helped me identify relevant factors. I decided how they are weighted and why.
AI suggested factors and proposed weighting them equally. Rejected — equal weighting ignores that some factors are more time-sensitive than others. I made all weight decisions independently:
| Factor | My Weight | My Reasoning |
|---|---|---|
| Days until test | 40% | Proximity to a deadline creates real pressure. This should dominate. |
| Topic confidence | 25% | Reflects actual understanding — the most direct measure of being prepared. |
| Credit value | 20% | Higher credit subjects carry bigger academic consequences if failed. |
| Grade risk | 15% | Useful signal, but less immediate than time and confidence. |
AI originally described the time component using the phrase "exponential decay." Rejected — I corrected this to urgency bands with sharply increasing risk near the test date. It is more honest and easier to explain and defend.
I chose insertion sort after asking AI to compare options for this use case.
AI confirmed insertion sort is appropriate for small datasets, is stable, and produces a comparator function that is easy to follow. I made two additional decisions independently:
Every sort decision is traceable back to the urgency score, which is traceable back to the four-component formula.
AI suggested keeping the UI minimal to reduce complexity. Rejected. GUI look and feel is 10 out of 30 available marks. A plain ranked list would not score near the maximum.
I directed the visual style. AI helped implement it. Key design decisions that were mine:
| Decision | Why I Made It |
|---|---|
| Red / amber / green risk system | Mirrors real-world risk dashboards. Readable without explanation. |
| Animated background orbs, orbit rings, particle effects | Raises visual quality. I specifically prompted for these elements and iterated on them. |
| Floating data nodes on the welcome screen | My idea — shows the type of data the app works with before the user even starts. |
| Score breakdown panel in deep dive | Makes the urgency formula visible to users and judges. Transparency by design. |
| Semester timeline | AI suggested it was optional. I kept it — it adds real value to the dashboard view. |
| Study tips ticker | My addition. Adds life to the interface and gives the app a sense of purpose beyond just data. |
I also used AI heavily during the styling phase, prompting specifically for visual details like the gradient text effects, the animated border on the recommendation box, the corner tick marks, and the geometric accent shapes. I reviewed every piece of CSS and adjusted values until the result matched what I had in mind.
Being specific about this matters. AI did not:
In several specific cases I directly overruled AI suggestions:
The urgency score formula for reference:
I used AI as a tool to think better, not to replace thinking. I asked questions, challenged suggestions, rejected ideas that did not fit the brief, and made all final decisions myself.
The urgency formula weights, the three-view structure, the visual design direction, and the decision to write insertion sort by hand were all my own choices. AI helped me explore options at each stage — but I decided what stayed and what did not.
This project reflects how I think about a real problem, built with AI as a tool, not as a replacement for judgment.