LoadBearing

An AI-powered family care coordinator using Amazon Bedrock and MCP to route care tasks across time zones, protect private moments, and visualize shared responsibility on a 3D globe.

Comment

Inspiration

Across modern diaspora families, millions of adult children live in different corners of the world while caring for aging parents back home. In these setups, caregiving invisible labor almost always collapses onto one person (usually the nearest local daughter or sibling), creating severe burnout, quiet resentment, and loneliness. When a sudden 3:00 AM doorstep alert or medical refill notice arrives, panic sets in. Yet, while it is 3:00 AM in Toronto, it is mid-afternoon in Pune or early evening in Berlin. We asked ourselves: What if we could use Amazon Alexa+ and Amazon Bedrock to run family caregiving like a resilient, 24/7 "Follow-the-Sun" system? That was the spark for LoadBearing — an intelligent, voice-first care-sharing platform that balances emotional and logistical caregiving fairly across time zones, protecting caregiver wellbeing while keeping parents safe and loved.

What it does

LoadBearing is an ambient care-coordination platform connecting aging recipients (like Dadi in Toronto) with global caregivers (Priya in Pune, Arjun in Berlin, Meera in Toronto) via Alexa+ and an interactive 3D Twilight Earth replica. Key features include:

  1. Follow-the-Sun Intelligent Handoffs: When an urgent doorstep event, prescription refill, or appointment occurs, LoadBearing's engine evaluates each caregiver's current local time, sleep/awake windows, and historical load share. It autonomously routes tasks to whoever is awake and least loaded.
  2. Doorstep Camera AI Triage: Ingests simulated Ring doorbell camera events. Regular courier deliveries are automatically logged as informational moments, while unexpected visitors or heavy deliveries require local human presence and route seamlessly to the right family member.
  3. Simulated Alexa+ Voice Surface: Family members interact naturally via Alexa+ ("Alexa, ask LoadBearing to hand off morning medicine check to Priya").
  4. Interactive 3D Photorealistic Earth Replica: A 3D globe powered by Three.js and NASA satellite imagery that displays the real-time day/night twilight terminator, glowing city lights, dynamic cloud cover, and caregiver pillars showing live awake/sleep states and emotional load distribution.
  5. Dignity & Privacy-First Consent Engine: Recipient consent rules, caregiver load-sharing visibility toggles, and private care moments protect family dignity.
  6. Global Accessibility & i18n: Full internationalization across 6 languages (English, Hindi, Spanish, German, French, Japanese) with Dark, Light, and Custom accessibility themes.

How we built it

We architected LoadBearing as a full-stack, enterprise-grade monorepo:

  • AI Agentic Layer & Amazon Bedrock: Powered by the Model Context Protocol (MCP), connecting Alexa+ and Bedrock agents to stateful tools (get_household_state, route_doorstep_event, submit_handoff_proposal, resolve_handoff).
  • Frontend Experience: Built with Next.js 16 (App Router & Turbopack), TypeScript, and Tailwind CSS.
  • 3D Earth Digital Twin: Built with Three.js and React Three Fiber, featuring:
    • NASA Blue Marble satellite day surface
    • Populated night-lights emission map
    • Realistic ocean specular sun-glint reflection
    • Atmospheric Rayleigh scattering glow
    • Animated, drifting 2048x1024 cloud layer
  • Timezone & Load-Balancing Engine: Built with Luxon, executing timezone math across India Standard Time (IST), Central European Time (CET), and Eastern Standard Time (EST) with simulated accelerated time controls (1x, 60x, 600x).
  • State & Data Store: Real-time optimistic state management via Zustand with JSON transactional storage and conflict versioning.

Challenges we ran into

  • Precision Geocoding on a 3D Sphere: Aligning Cartesian Three.js sphere vertices and UV coordinates with spherical latitude and longitude (Pune at 18.52° N, 73.86° E, Berlin at 52.52° N, Toronto at 43.65° N) so that pillar bases and handoff arcs accurately anchor onto real land masses.
  • Realistic Day/Night Twilight Shader: Designing a custom WebGL GLSL shader that computes dynamic sun vectors based on real-world UTC timestamps, blending daytime satellite land, city lights on the dark hemisphere, and a warm golden sunset terminator without harsh seams or performance degradation.
  • High Contrast in Extreme Themes: Supporting high-contrast light mode, deep dark mode, and custom accessibility modes across translucent frosted glass surfaces without losing WCAG AA readability.

Accomplishments that we're proud of

  • Flawless Follow-the-Sun Routing: Watching an urgent doorbell event seamlessly re-route halfway across the planet to an awake sibling rather than waking up a sleeping family member.
  • Authentic 3D Earth Replica: Creating a photorealistic, performant Earth globe with 60 FPS rotation, cloud drift, and live sunlight reflections.
  • Production-Grade MCP Integration: Implementing the Model Context Protocol (MCP) server ready for Bedrock-powered Alexa+ agents.
  • Inclusive & Global: 6 international languages supported out of the box with instant switching.

What we learned

  • How to design compassionate, human-centric AI interactions that prevent caregiver burnout without reducing human care to cold tickets or tasks.
  • Techniques for building real-time shaders in WebGL and React Three Fiber that balance visual fidelity with lightweight web performance.
  • Implementing the Model Context Protocol (MCP) to bridge LLM reasoning with domain-specific stateful tools.

What's next for LoadBearing

  • Physical Alexa Skill Deployment: Publishing the companion Alexa+ Skill to the Amazon Alexa Skill Store with proactive ambient notifications on Echo Show devices.
  • Direct Ring & Blink Doorbell API Webhooks: Connecting live Ring video doorbells to auto-classify motion with Amazon Bedrock Multimodal vision.
  • Care Budgeting & Micro-Reimbursements: Integrating Amazon Pay for family members to automatically split pharmacy orders, groceries, and medical equipment seamlessly.

Built With

Share this project:

Updates

Submission history