Package Whisperer

Dignified elder care for Ring. Replaces noisy chime alarms with calm delivery windows and warm, empathetic check-in digests powered by Amazon Bedrock.

Comment

Inspiration

Smart video doorbells are incredible pieces of technology, but if you've ever watched an elderly family member live with one, you quickly notice something frustrating: they are ridiculously loud, intrusive, and anxiety-inducing.

Every time a courier drops a parcel on the porch:

  1. The in-home chime blares: The resident jumps, feels startled, or rushes to the door before they've even put their slippers on.
  2. The remote caregiver's phone buzzes: A son or daughter gets an urgent alert 100 miles away. They immediately worry ("Did mom fall trying to pick it up?") and call immediately—inadvertently stripping their parent of independence over a simple cardboard box.

I built Package Whisperer for the Ring Track of the Amazon Developer Hackathon because good caretaking shouldn't yell—it should whisper.

What It Does

Package Whisperer transforms standard Ring doorbells into a calm, dignified eldercare guardian:

  • Silent Delivery Grace Period: When a courier drop-off is detected, instead of blaring loud sirens, the system starts a peaceful 4-hour window (or 30s in demo mode).
  • Ambient Resident Calm Tablet (/resident): Displays a soothing, low-stimulus screen ("A delivery is resting on the porch") with soft Web Audio chords and a single-tap "Got it, thank you!" confirmation button.
  • Independence Scoring: Tracks and celebrates how often residents retrieve their own packages without requiring caregiver intervention.
  • Quiet Hours (10 PM – 7 AM): Pauses non-critical notifications overnight so neither seniors nor family members are woken up.
  • Empathetic Amazon Bedrock Escalation: If a package remains uncollected after the grace period, it invokes Amazon Bedrock (Claude Haiku 4.5) to compose a warm, human check-in note (customizable into Gentle, Direct, or Urgent tones) instead of a scary red alarm.

How I Built It

Package Whisperer is built as a real-time event pipeline:

  1. Ring Webhook Ingestion: Ingests live button_press and motion_detected events, verified via cryptographic HMAC-SHA256 signatures (X-Signature) to guarantee authenticity.
  2. Heuristic Motion Classifier: Automatically filters out neighborhood dogs, cats, and passersby by inspecting Ring event duration and human/motion classification flags.
  3. State Machine & Timer Scheduler: Built with Node.js (v22) and Express 5.x, managing delivery lifecycles, quiet-hour pauses, and grace periods.
  4. Generative Caretaker Digests: Leverages @aws-sdk/client-bedrock-runtime to query Claude Haiku 4.5 on Amazon Bedrock with strict formatting constraints, producing ~300ms empathetic notifications.
  5. Three Web Interfaces:
    • Caretaker Hub (/caretaker): Real-time monitoring dashboard, countdown timers, tone selector, and independence metrics.
    • Resident Calm Screen (/resident): High-contrast, senior-friendly ambient UI using the browser Web Audio API for harmonic acoustic feedback.
    • Interactive Simulator (/simulator): Live testing sandbox for judges to simulate rings, deliveries, pickups, and fast-forward timers.

Challenges I Ran Into

  • Defensive Webhook Parsing: Depending on the Ring device hardware, webhook property schemas can vary (e.g., top-level device_id vs. nested attributes). I had to build a robust normalization layer to ensure payload consistency.
  • Bedrock Throttling & Offline Resilience: If AWS token quotas are reached or internet drops, eldercare systems must never crash. I implemented a zero-dependency local tone engine that seamlessly takes over if Bedrock calls face latency or throttling exceptions.
  • Eliminating False Alarms: Balancing prompt courier drop-off recognition against harmless porch movement required fine-tuning visit window durations and motion tags.

Accomplishments That I'm Proud Of

  • End-to-End HMAC Security: Full cryptographic webhook verification with constant-time comparisons to prevent timing attacks and zero hardcoded secrets.
  • Sub-400ms AI Latency: Blazing-fast Bedrock inference that turns cold sensor data into genuinely warm, caring messages.
  • Dignity-First Design: Creating a smart home technology experience that actively respects senior autonomy rather than treating them like patients under surveillance.

What I Learned

  • How to integrate and authenticate real-time webhooks from the Ring developer platform.
  • Best practices for structuring low-latency inference calls using the AWS Bedrock SDK and Claude Haiku 4.5.
  • The importance of "calm computing" principles when designing technology for older adults.

What's Next for Package Whisperer

  • Weather-Aware Escalation: Shortening the grace period automatically if rain, snow, or extreme heat is detected via local weather APIs.
  • Two-Way Ring Intercom Integration: Allowing caregivers to send Bedrock-synthesized voice whispers directly through the Ring chime speaker at scheduled times.
  • Multi-Resident Support: Expanding the dashboard for assisted living facilities and senior community coordinators.

Built With

Share this project:

Updates

Submission history