Inspiration
Since we were all students, we listed out our weaknesses so we could help other students. Together, we found four main problems we all needed help on.
- Collaborating with people we don't know was hard.
- We consistently had writer's block, so writing stories was difficult.
- Studying hard subjects was difficult, especially since tutoring was expensive.
- Negotiating with parents or people in general.
What it does
Luminary is a student hub built from students for students. Luminary has four main features, StoryWeaver, StudyFlow, TutorHub, and Negotiator.
StoryWeaver is a program where students can create stories with the help of AI. Each new story gets a code that can be shared to other users, allowing students to collaborate on each other's stories.
StudyFlow is a program in which students enter what they would like to study and AI generates questions for the student to practice on. The AI also explains the answer and you can ask it questions. StudyFlow also keeps track of your score percentage.
TutorHub allows users to sign up as tutors or contact other tutors to ask for tutoring. When a user contacts a tutor, the tutor receives an email and a notification on their profile. They can then respond to this email.
Negotiator is a program that teaches students how to negotiate with their parents. Simple type in what you want, what your parents' most likely objection is, and what you will compromise. An AI then proposes the three most likely ways for you to convince your parents, alongside reasoning for why this can convince them. There is also a simulator on the side that allows you to simulate this conversation to your parents.
However, these are not the only functions Luminary has! Mochi, a cute cat in the corner, is an AI powered chatbot that allows students to talk to and explain any emotional problems they want Mochi to comfort them with. This cat also sleeps when not interacted with for a minute and instantly wakes up when your mouse hovers over her!
How we built it
We started off by grouping together and discussing plans and ideas on our shared Google doc. After our idea was created, we inputted a prompt into Claude and Gemini to create the first prototype. Our backend was all created with Python. By version 4, we were adding in a Gemini API so we could use Gemini 2.5 Flash on our project (this can be replaced by an AI API, for example Dora was able to make it work with OpenAI API). At this point, it was still a local only website. By version 5, we had a multi-threaded HTTP server to handle multiple users at once and still keep it fast. Our database was created with SQLite so that student stories, tutors, and messages were all saved securely. The frontend was crafted with HTML5, CSS3, and JavaScript. Our sleep or wake feature for Mochi used JS timers. We finally moved this onto a real website with a domain by uploading Luminary to GitHub and using Render to put it on a domain.
Challenges we ran into
We ran into and overcame many challenges while building Luminary. For example, the first three versions were all prewritten and precoded, having no AI whatsoever. This was soon changed when we added an API into version 4. We sometimes stayed up until 2 AM just to finish coding the majority of Luminary. We panicked around 13 hours before the project was due because we had no way of transporting an local document online, but that was solved by using GitHub and Render.
Accomplishments that we're proud of
We were very proud of the entire project, especially when the API got implemented. All four sections of this project worked so well, we couldn't help but be so excited we got yelled at by our parents for screaming too loud. We are especially proud of the transition from local to online as we have never tried that before. Mochi being able to sleep was also an accomplishment we were proud of as it just made the entire website feel more alive.
What we learned
We gained hands-on experience with Full-Stack development, learned how a Python backend communicates with a JavaScript frontend, and mastered Prompt Engineering, like figuring out how to give AI specific "personalities" like a skeptical parent or a supportive cat. We also learned how to transfer a local website to an online one, how to implement a Gemini API into our code, and how to design a website to have all the features we like.
What's next for Luminary
We want Luminary to be even more helpful for students. We plan to:
- Give Mochi a real voice so she can talk back to students who are stressed.
- Add profiles to users so they can describe themselves and allow others to easily contact them.
- Add a streak to StudyFlow so students are more motivated to study.
- Add a place for you to rate and see the rating of tutors.
- Give Mochi a body so she can walk around on the screen and be more interactive.
Log in or sign up for Devpost to join the conversation.