NovaLabs

## About the Project

Our project was inspired by a simple problem we see all the time at UT Dallas: students have access to a lot of useful campus information, but it is spread across too many places and is not always easy to use in the moment. A student might need to find a professor, check room availability, figure out where to go for help, or make use of time between classes, but doing that usually means opening multiple tabs and searching through different systems. We wanted to build something that felt more natural and actually useful for student life.

That led us to create an AI-powered campus help desk for UTD. The idea was to let students type questions in plain language, like they normally would, and get back clear, relevant, campus-specific help. Instead of making students adapt to the system, we wanted the system to adapt to the student.

What especially pushed our idea forward was combining two directions: real campus data and AI-based text understanding. We used the Nebula API as the backbone for structured UTD data such as courses, professors, grades, rooms, and events. On top of that, we designed an NLP-driven help flow that could interpret student intent and route them toward the right answer or feature. That is how the project became more than a chatbot. It became a student support tool.

Some of the main features we focused on were **Professor Pick**, **Who Do I Even Ask?**, **Dead Time Optimizer**, and the **Student Heat Map**. Professor Pick helps students explore courses and professors more easily. Who Do I Even Ask? is meant to reduce confusion by guiding students toward the right resource or next step. Dead Time Optimizer helps students make better use of awkward gaps between classes. Student Heat Map takes a broader view by surfacing common student pain points and showing where friction happens across campus.

### How We Built It

We approached the project as both a design and systems problem.

On the frontend side, we designed a futuristic interface centered around a main search bar, with feature shortcuts and a sidebar inspired by modern AI products. We wanted the UI to feel familiar, clean, and easy to use while still standing out visually. The homepage was built to immediately show the student what they could ask, and the results flow was designed to return insights and next steps rather than generic output.

On the backend and data side, we planned around the Nebula API endpoints that were most relevant to our goals. That included endpoints for:

- course data
- professor data
- grade information
- rooms and scheduling
- campus events and calendar data

These data sources allowed our features to feel grounded in real UTD information instead of being purely hypothetical.

For the AI side, our main focus was on taking a student’s natural-language question and identifying what kind of help they actually needed. In simple terms, the flow was:

1. student enters a query  
2. system detects the likely intent  
3. relevant campus data is pulled  
4. results are organized into insights and actions  

So a question like *“Find me a quiet room near ECSW before my 4 PM class”* is not treated like a generic text prompt. It becomes a room-finding task with extracted details like location, time, and preference.

### Challenges We Faced

One of the biggest challenges was making the project feel both ambitious and realistic. We had a lot of ideas early on, but not every idea could be built at full depth in a hackathon setting. We had to keep asking ourselves what would be most useful to students and most demoable to judges.

Another challenge was figuring out how to make the AI layer actually meaningful. It is easy to say a project uses AI, but much harder to make it feel useful instead of gimmicky. We had to think carefully about how natural language input should map to real campus actions, and how to make the results look relevant rather than generic.

We also ran into product design challenges. Since our platform includes multiple help modes, we needed a layout that felt simple instead of overwhelming. That is why we centered the design around a main search experience, then supported it with focused feature modules.

### What We Learned

We learned that building a good project is not just about adding features. It is about connecting data, design, and user need in a way that feels clear. We also learned how much stronger a project becomes when it solves a real, local problem instead of being overly broad.

Technically, we learned how to think about API-driven app design, feature scoping, search-based UX, and how NLP can be used as a routing and decision-support layer rather than just a text generator. We also learned that presentation matters a lot: when the problem, flow, and impact are easy to understand, the project becomes much stronger.

### Why This Project Matters

At its core, this project is about reducing friction in student life. UTD students already have a lot on their plates, and even small inefficiencies add up. We wanted to build something that could save time, reduce confusion, and make campus resources feel easier to access.

Our goal was not just to make another student app. It was to build something that turns campus questions into useful next steps and makes the student experience feel a little more connected, efficient, and human.

Built With

• css
• figma
• html
• javascript
• react