BloomCube AI

BloomCube AI turns office CCTV cameras into a wellness coach, reading posture and well-being cues to gently nudge healthier habits at work!

Comment

💡 Inspiration

As corporate employees ourselves, we know what it feels like to lose ourselves in deep work, deadlines, and constant pressure. There were so many days when we'd sit glued to our chairs for hours without moving, skipping hydration and lunch, and ending up in terrible posture at 5 PM. It wasn't just "bad habits", it genuinely affected our mental clarity, energy, and mood.

When we started researching, we realized we weren't alone. Studies show that employees spend over one-third of their entire year inside their workplace. If a place consumes that much of our life, shouldn't it care about our well-being?

The spark came from a simple insight: What if the same office cameras already installed for security could quietly help us live healthier, without judging, tracking, or spying?

That thought merged with this hackathon theme, stitching together computer vision, wellness psychology, AI automation, and real-time Slack nudges to create BloomCube AI, a privacy-safe, AI-powered wellness layer that turns passive CCTV into active ergonomic intelligence.

🚀 What It Does

BloomCube AI transforms existing office CCTV cameras into a real-time wellness monitoring and intervention system. It detects posture issues, prolonged stillness, fatigue patterns, and cold/shivering movements, then sends friendly, non-intrusive nudges via friendly Slack bot named Bloomy to encourage healthier behaviours and notify admins for environmental adjustments.

Core Workflow:

  1. Admin connects CCTV footage → System maps cubicles from a single camera angle
  2. AI analysis pipeline runs: OpenCV + MediaPipe extracts pose landmarks; proprietary scoring engine detects poor posture, fatigue, stillness, and cold/shivering movements
  3. Rules engine + AI nudges fire → Slack bot sends real-time wellness alerts ("Stand up & stretch!", "Take a hydration break!") AND notifies admins of temperature concerns ("Employee at Cubicle 12 is cold, adjust HVAC")
  4. Dual dashboards track trends: Admins see anonymized wellness heatmaps by cubicle/team; employees see personal wellness history and nudge impact

System Architecture Visualization: System Architecture Visualization Flowchart This integrated pipeline demonstrates:

  • End-to-end data processing without raw video storage (privacy-first)
  • Parallel paths for admin notifications (temperature alerts) and employee wellness nudges
  • Real-time processing pipeline with multiple detection layers
  • Dual-dashboard design showing different insights to different stakeholders
  • Scalability: handles multi-camera, multi-cubicle scenarios

Why It's Different:

  • Privacy-by-Design: Only pose landmarks are stored; raw video is discarded immediately
  • Zero New Hardware: Leverages existing security cameras, no new infrastructure, no extra power consumption
  • Workplace-Wide Impact: Generates office-level wellness heatmaps and team analytics, not just individual surveillance
  • Intelligent Alerting: Dual-layer rules (hard thresholds) + AI-assisted nudge logic (learned patterns) for smarter interventions
  • Environmental Awareness: Detects cold/shivering and alerts admins for comfort optimization (can later integrate with HVAC systems)
  • Built in Hackathon Timeframe: Entire full-stack system shipped rapidly using AI-assisted development

🛠️ How We Built It

Backend (FastAPI + PostgreSQL):

  • RESTful APIs for video upload, cubicle mapping, posture scoring, cold detection, and wellness analytics
  • JWT authentication (python-jose + passlib/bcrypt) for secure admin & employee portals
  • Real-time Slack SDK integration for instant nudge delivery and admin notifications
  • Alembic for database migrations; SQLAlchemy ORM for schema management

Computer Vision/AI Pipeline (Python):

  • OpenCV for video decoding and frame extraction
  • MediaPipe Pose for real-time pose landmark detection (33-point body model)
  • Custom scoring algorithms: posture quality (spine angle, shoulder alignment), fatigue markers (head position, blink patterns), movement tracking, and cold/shivering detection (trembling motion patterns)
  • Slack Nudges personalised and generated via AI

Frontend (React + TypeScript):

  • Admin dashboard: CCTV frame viewer, cubicle mapper, wellness analytics (heatmaps, trends by role/team), temperature alerts [Demo Admin: Email - [email protected], Password - admin123]
  • Employee dashboard: Personal wellness trends, nudge history, trend insights [Demo Employee: Email - [email protected], Password - demo123]
  • Clean, responsive UI with color-coded wellness scores (green=healthy, yellow=caution, red=intervention needed)

Deployment:

  • Render: Backend (FastAPI with auto-scaling), Database (PostgreSQL with secure, encrypted fields for sensitive data)
  • Vercel: Frontend

🎨 Design & Aesthetics

UX Philosophy: Wellness alerts should feel like friendly coaching, never like surveillance or punishment.

  • Minimal, clean dashboards with intuitive wellness metrics and historical trends
  • Real-time notifications via Slack (the tool employees already use) for seamless integration
  • Color-coded wellness indicators for instant visual feedback on team/individual status
  • Temperature alerts for admins showing which cubicles need climate adjustments
  • Responsive design across desktop, tablet, and mobile, works everywhere employees access it

✨ Key Technical Achievements

✓ Full pipeline implemented and live: Video → Pose Detection → Scoring → Cold Detection → Rules/AI → Slack → Dashboards
✓ Live demo and hosted systems: Frontend (Vercel), Backend API (Render with /docs), Database (PostgreSQL)
✓ Multi-user auth: Separate role-based access for admins and employees
✓ Real-time integration: Slack bot with configurable wellness and temperature alert triggers
✓ Cold/Shivering Detection: AI-powered trembling motion detection for employee comfort monitoring
✓ Cubicle mapping: Automatic spatial detection from single CCTV camera angle
✓ Anonymized analytics: No PII stored; team-level aggregations preserve privacy while enabling data-driven insights

🌱 Sustainability & Climate Impact

BloomCube AI contributes to environmental sustainability through reduced workplace burnout, attrition, and energy waste:

  • Lower Burnout & Attrition: Healthier employees = fewer resignations = reduced need for constant hiring, onboarding, and resource wastage from high turnover
  • Fewer Unnecessary Trips: Better ergonomics and fewer health issues reduce commutes to clinics, therapy, and medication dependence
  • Energy-Efficient: Software-only wellness layer using existing cameras, zero new hardware, zero additional power consumption
  • Workplace Efficiency: Healthier, more alert employees = higher productivity per hour = lower resource and energy usage per unit of productive output
  • Temperature Optimization Foundation: Cold detection data provides the intelligence layer for future HVAC automation (Phase 2), enabling significant energy savings

At scale across large organizations, this translates to measurable reductions in absenteeism, healthcare costs, and carbon footprint per employee.

📚 What We Learned

  • Privacy is paramount: Employees genuinely want wellness nudges only if they trust the system doesn't spy on them
  • Integration is key: Real wellness adoption requires meeting employees where they already work (Slack/Teams, not a new app)
  • Environmental awareness matters: Detecting comfort issues (cold/warmth) opens doors to building automation and energy optimization
  • Modularity matters: Building clean, composable CV pipelines, rules engines, and nudge logic makes iteration fast
  • Multi-tech harmony: Stitching computer vision, rules engines, AI, and Slack APIs together requires careful API design and error handling
  • Hackathon power: Using AI-assisted development framework dramatically accelerated shipping of a full-stack system in limited time

🎯 What's Next for BloomCube AI

HealthTech & Wellbeing:

  • Integration with employee assistance programs (EAPs) and HR wellness initiatives
  • Anonymized wellness insights for company-wide policy changes (e.g., standing desk adoption, micro-break norms)
  • Personalized coaching profiles based on detected patterns

AI/ML Enhancements:

  • Learned fatigue models using micro-expressions and eye-tracking (stress, sleepiness)
  • Anomaly detection in wellness patterns (sudden degradation signals burnout risk)
  • Nudge optimization: ML-driven personalization of when and how to alert each employee for max adoption

Sustainability & Scale:

  • Pilot programs with 3-5 major MNCs across their office campuses
  • Measure and correlate wellness improvements with energy usage, absenteeism reduction, and retention metrics
  • Public dashboards showing ROI on workplace wellness investment (both health and environmental metrics)

Phase 2: BloomCube Facilities (Smart HVAC & Building Optimization)

  • IoT Integration: Connect cold detection signals with building HVAC systems (Honeywell, Johnson Controls, etc.)
  • Occupancy-Based Climate Control: Automatically adjust temperature when employees are away or floor sections are empty (occupancy detected via CCTV + motion patterns)
  • Energy Cost Optimization: Reduce HVAC runtime during periods of low occupancy; target 10-15% energy savings on cooling costs
  • Smart Zone Management: Create micro-climate zones based on cubicle usage patterns; reduce cooling power when detecting departures (bag pickup, prolonged absence)
  • Sustainability Metrics: Real-time energy savings dashboard; ROI tracking for energy cost reduction per building
  • Measurable Impact: For a 500K sq ft office spending $1-1.5M/year on HVAC, potential savings of $100-150K/year with 2-year payback on software investment

Extended Reach:

  • Live camera support (beyond video uploads) for real-time, always-on wellness monitoring
  • Multi-camera calibration for large open offices and bullpens
  • Emotion + micro-expression detection for deeper stress and fatigue insights

🔮 Long-Term Vision

"Create workplaces where well-being isn't an afterthought, it's built into the environment."

We believe BloomCube AI is the first step toward AI-powered, privacy-safe workspaces that respect employee autonomy while genuinely improving their health, happiness, and productivity. By 2026, we envision this running across major tech and corporate campuses globally, reducing burnout, improving retention, and proving that AI can be a force for human wellness, not surveillance. With Phase 2 (BloomCube Facilities), we'll extend this to intelligent building automation, creating sustainable, employee-centric offices that optimize both human and environmental well-being.

Built With

Share this project:

Updates