Control+Plan

Inspiration

Our project focuses on balancing academic responsibilities with social life of university students. As students ourselves, we understand this struggle firsthand and wanted to create a tool that makes planning easier and more adaptive.

What it does

Control+Plan is designed to rapidly adapt and update timelines as student received new information about events and social hangouts. It helps students efficiently manage their schedules, ensuring that both social events and academic deadlines are seamlessly reflected.

How we built it

We built Control+Plan using a modern full-stack approach. Our tech stack includes Supabase for the backend database, TypeScript for strong typing, React/Next.js for the frontend development and rendering, and Gemini for AI-driven planning capabilities.

Challenges we ran into

One major challenge was the limited resources provided by the free LLM model we used, which made processing slower and lacks fine-tuning (RAG) options. Additionally, the limited timeframe of the hackathon also meant we could not develop a bespoke learning model in place of the free model that we had.

Accomplishments that we're proud of

We successfully integrated across the full stack, connecting the user interface to the backend relational database. Our team collaborated effectively by allocating tasks based on individual strengths. Seeing all sections finally integrated together to produce a working prototype for the first time was a huge milestone and a proud moment for our team.

What we learned

We learned the value of rapid prototyping and how to prioritize features under tight deadlines. Working with AI models taught us practical considerations of resource constraints, while integrating front-end and back-end systems strengthened our full-stack development skills. Most importantly, we realized the power of thorough planning with focus on user-centric specification.

What's next for Control+Plan

Looking ahead, we would love to attempt developing a bespoke AI model from the ground up. This model will be specifically optimised for planning and producing structured outputs, rather than focusing on general NLP capabilities. We want to design a model that excels at creating actionable schedules, adapting to changes efficiently, and supporting decision-making for university students. This will allow Control+Plan to move beyond a prototype and become a truly intelligent planning assistant tailored to users’ needs.