focusBuddy

nspiration

Productivity apps rely on willpower; focusBuddy uses computer vision to tell the truth about your attention.

What it does

Tracks your focus in real time with your webcam, logs sessions, and gives you an AI-generated daily plan — all fully privacy-safe.

How we built it

MediaPipe for face tracking, COCO-SSD for phone detection, Skyflow for secure data vaulting, Sanity for session summaries, Parallel for the AI coach.

Challenges

Smooth real-time CV, strict privacy rules, syncing three external systems without blowing up the architecture.

Accomplishments

Real-time focus detection, zero image storage, tokenized user data, clean data pipeline, and a genuinely helpful AI study coach.

What we learned

Lightweight CV can be both fast and privacy-preserving; Parallel agents shine when given structured behavioral data.

What’s next

Mobile version, smarter distraction predictions, browser blocking, and more personalized coaching.

Built With

• and-analytics.-parallel-ai-?-agent-generating-personalized-daily-focus-plans.-html/css/canvas-?-ui
• animation-loop
• blink-detection.-tensorflow.js-(coco-ssd)-?-phone-detection-directly-in-the-browser.-skyflow-?-zero-trust-privacy-vault-for-sensitive-session-data.-sanity-?-structured-backend-for-session-summaries
• built-with-javascript-+-vite-?-lightweight-frontend-powering-real-time-detection-at-30?60-fps.-mediapipe-face-mesh-?-face-landmarks
• for
• head-pose-estimation
• local
• real-time-graphs
• storage
• streaks
• timelines.-web-apis-?-webcam-access
• tsc
• user