FocusFlow — Project Story
Inspiration
As a high school student, I've noticed that spending hours studying doesn't necessarily mean spending hours focused. It's easy to start a study session with good intentions, only to get distracted by another browser tab or lose concentration without realizing it.
Most productivity apps measure how long you study, but I wanted to explore something different: What if a study app could help you understand how consistently you're paying attention?
That question inspired FocusFlow.
What It Does
FocusFlow is an AI-powered study companion that combines real-time gaze tracking, structured study sessions, and website blocking.
Using Google's MediaPipe Face Landmarker, FocusFlow analyzes facial landmarks through a webcam to estimate whether a student is looking toward their screen or looking away.
Before starting a session, users complete a nine-point calibration process that adapts gaze estimation to their setup.
During a study session, FocusFlow provides live attention-related feedback. Afterward, students can review their session history and analytics to better understand their study habits.
I also built FocusFlow Distraction Shield, a companion Chrome extension that blocks user-selected distracting websites. Students can configure their blocklist and control when website blocking is active.
Together, these features help students become more intentional about how they spend their study time.
How I Built It
I developed FocusFlow using Next.js, React, TypeScript, and Tailwind CSS.
For the computer vision component, I integrated Google's MediaPipe Face Landmarker and built a calibration workflow to support gaze estimation.
The application combines webcam-based tracking with a focus timer, attention-related state management, and session analytics.
For website blocking, I developed a Chrome Manifest V3 extension using the declarativeNetRequest API. The extension communicates with the FocusFlow web application to coordinate blocking with study sessions.
I used AI-assisted development tools, including OpenAI Codex and Freebuff, during development and UI exploration.
Challenges I Ran Into
One of the biggest challenges was making real-time gaze tracking reliable without slowing down the application.
Processing webcam frames can be computationally expensive, and interruptions in inference can affect calibration and live tracking.
Another challenge was handling situations where the user switches browser tabs. Browsers may throttle background processing, which makes continuous gaze estimation unreliable.
Building the Chrome extension introduced additional complexity because the extension and website operate separately. I had to account for synchronization, browser permissions, and blocking rules.
These challenges taught me that integrating an AI model is only one part of building a usable AI application. Performance, reliability, and user experience are equally important.
Accomplishments I'm Proud Of
I'm particularly proud of bringing multiple components together into one application.
FocusFlow goes beyond a basic study timer by combining AI-assisted gaze estimation, personalized calibration, focus analytics, and browser-level distraction blocking.
I'm also proud of building a companion Chrome extension rather than limiting the project to features available inside a website.
The result is a study tool that addresses distractions from two directions: providing feedback about visual attention and reducing access to distracting websites.
What I Learned
Building FocusFlow helped me understand how pretrained computer vision models can be integrated into real-world applications.
I gained experience with facial landmark detection, calibration workflows, real-time browser processing, Chrome extension development, and communication between separate application components.
I also learned the importance of recognizing AI limitations. Looking at a screen doesn't necessarily mean someone is concentrating, so FocusFlow should be understood as a visual-attention estimation tool rather than a perfect measure of focus.
What's Next for FocusFlow
I want to improve the efficiency and accuracy of gaze estimation, make calibration more reliable, and expand the analytics available to students.
I'd also like to make Distraction Shield easier to install and test across different browsers and devices.
In the future, I hope to evaluate FocusFlow with students to understand whether its feedback and distraction-management features help them develop more consistent study habits.
Ultimately, my goal is to make FocusFlow a tool that helps students study more intentionally, not simply study longer.
Built With
- chrome-extensions-api
- css
- html
- javascript
- mediapipe
- next.js
- react
- tailwind-css
- typescript
- vercel