FocusAI
Inspiration
I had this idea when I caught myself doomscrolling. I was feeling intimidated by the hackathon and didn't have any concrete ideas, so I was just wasting time on my phone. Then it hit me like lightning: WAIT, THIS IS IT! My phone is literally a barrier to learning, and I know I'm not alone in this struggle. So many students use their phones while studying without even realizing it. That moment of self-awareness became the foundation for FocusAI.
What it does
FocusAI is an AI-powered accountability partner that tackles phone addiction head-on during study sessions. Using computer vision and intelligent tracking, it creates a comprehensive learning environment that detects when phones enter your study space in real-time. FocusAI isn't just another productivity app that blends in, it's a complete productivity suite that includes focus timers, task management, AI-powered text summarization, and beautiful analytics to track your progress. Think of it as having a smart study buddy that keeps you accountable while providing all the tools you need in one aesthetic, distraction-free platform.
How I built it
I built FocusAI using Python and Django in the backend, with TensorFlow and OpenCV for the computer vision. The phone detection uses a custom AI model trained with Google's Teachable Machine, which required careful dataset balancing to avoid false positives. The frontend combines HTML, CSS, and JavaScript to create a modern, glassmorphism-inspired interface that appeals to dopamine-addicted Gen Z users while staying simple enough for millennials. I integrated HuggingFace Transformers for the AI text summarization feature using the DistilBART model, and built a comprehensive database system to track focus sessions, tasks, and user progress over time.
Challenges I ran into
Where do I even start with this one?! The AI model was throwing false positives left and right until I figured out I needed to balance the dataset better and reduce the training epochs. My camera decided to stop cooperating at the worst possible moment, and integrating Python with JavaScript through Django felt like learning a completely new language. I was literally about to give up due to the database management, it was SO complicated and made me question all my life choices. I CAN NOT trust AI with a database, but eventually, after annoying Claude over and over again by asking it to explain, I somehow managed to pull it off.
Accomplishments that I'm proud of
I'm incredibly proud that this is my first hackathon and I actually built something that works! Creating a fully functional AI-powered study tool with computer vision, real-time tracking, and multiple integrated features feels surreal. The aesthetic design turned out exactly how I envisioned, modern and engaging without being distracting. Most importantly, I solved a real problem that affects millions of students worldwide (and me), and I did it in a way that's accessible to anyone with a webcam. The fact that everything runs locally means users maintain complete privacy while getting powerful AI assistance.
What I learned
This hackathon taught me so much beyond just coding. I learned how Django really works under the hood, especially the complex parts I'd been avoiding. I discovered how to handle the intense stress and pressure of a hackathon environment, and got a crash course in what these events are actually like. But the biggest game-changer? The browser's "inspect" tool is an absolute LIFE SAVER, I don't know how I survived coding without it before, especially the frontend. I also learned that sometimes the best ideas come when you're procrastinating, and that building something you personally need makes the development process so much more meaningful.
What's next for FocusAI
This is just the beginning! Next up is improving the backend architecture and potentially migrating from the current database to PostgreSQL for better performance and scalability. I want to enhance the AI model to work across all types of phones and even detect other distracting devices like tablets or smartwatches. Future features include a mobile companion app for cross-device synchronization, advanced analytics with ML-powered study recommendations, integration with popular learning management systems, and collaborative focus sessions for study groups. The goal is to make FocusAI the ultimate inclusive learning platform that breaks down barriers to focused, productive studying for everyone.
Tools I used
Stitch.ai for the frontend, and some code snippets from Claude/ChatGPT for the backend of the timer logic. I also used Teachable Machine for the AI model. Claude for the GitHub ReadMe.
Log in or sign up for Devpost to join the conversation.