Aetheris Omni-Agent

Autonomous cross-device concierge unifying Alexa+, Ring, Bee AI & Fire TV via AWS Bedrock & Amazon MCP (Spec 2025-11-25) to proactively solve schedule conflicts, courier security & Prime restock.

Comment

Inspiration

Today’s smart homes and personal assistants are fundamentally reactive and fragmented: users must issue disconnected, single-turn voice commands to Alexa, manually check motion notifications on Ring, review biometrics on wearable devices, and navigate smart TVs.

We asked: What if our living space operated as a unified, proactive cognitive orchestra?

With the announcement of the Amazon Model Context Protocol (MCP Spec 2025-11-25) and AWS Bedrock Converse API, we saw an unprecedented opportunity to build Aetheris Omni-Agent—an autonomous ambient concierge that connects Alexa+, Ring Video Doorbell, Bee Wearable AI, and Fire TV into one self-orchestrating system that resolves conflicts before the user even has to ask.

What it does

Aetheris Omni-Agent actively monitors context across devices to autonomously coordinate life and work:

  1. Autonomous Schedule Collision Solver (Alexa+ Track):

    • Continuously evaluates calendar events via MCP tools (check_schedule_conflicts & reschedule_event).
    • Autonomously detects double-bookings, shifts overlapping sessions to open time slots, syncs changes to SQLite WAL storage, and immediately broadcasts the updated itinerary to Fire TV in 4K resolution.

  2. Ring Courier Verification & Amazon 1-Click Auto-Procurement (Cross-Device Symphony):

    • When Ring Doorbell detects an Amazon Logistics courier, Aetheris authenticates the courier credential and temporarily disengages the smart deadbolt for 60 seconds with an audit log.
    • Simultaneously checks office inventory (get_inventory_status), detects depleted coffee pods, and triggers an autonomous Amazon Corporate 1-Click restock order (auto_procure_supplies), displaying the animated Prime receipt on screen.

  3. Bee Wearable AI Biometrics -> Smart Lighting & Ambient Canvas:

    • Ingests real-time wrist biometrics (heart rate & cognitive stress index).
    • Upon detecting elevated stress (78%), it intervenes proactively: shifts room lighting to 5,500K focus daylight, mutes doorbell chimes, delivers wrist haptics, and sets Fire TV to a calming focus canvas.

  4. Official MCP Spec (2025-11-25) Protocol Inspector:

    • Fully open-standard JSON-RPC 2.0 streaming over Streamable HTTP and Server-Sent Events (SSE), enabling zero-latency tool execution and live multi-modal card rendering.

How we built it

  • Orchestration & AI Core: Python with FastAPI, implementing the official MCP 2025-11-25 specification over Streamable HTTP SSE (/mcp and /api/events).
  • AWS Bedrock Converse API: Integrated via boto3 utilizing Anthropic Claude 3.5 Sonnet and Amazon Nova, with an autonomous agent fallback loop.
  • Cross-Device State Hub: Real-time asynchronous state manager handling synchronized telemetry across Alexa+, Ring, Bee Wearable, and Fire TV.
  • Storage: SQLite database operating with WAL mode, powering 6 unified schemas (schedule, inventory, orders, ring_events, bee_telemetry, firetv_canvas).
  • Executive Frontend Simulator: Dark glassmorphic studio featuring a live pulsating Alexa Orb, real-time waveform visualizer, Web Speech API microphone recognition with live transcription, and collapsible MCP packet inspector.

Challenges we ran into

  • MCP Spec 2025-11-25 Keepalive & Stream Buffering: Maintaining persistent Server-Sent Events (SSE) connections through HTTP proxies required configuring periodic : keepalive\n\n pings.
  • AWS Bedrock Tool Schema Strictness: Bedrock’s Converse API enforces strict JSON Schema definitions (disallowing default keys inside schema properties). We engineered a schema sanitizer to guarantee 100% compliant payload structures.
  • Race-Free High-Frequency Procurement: High-speed automated 1-Click orders previously risked collision on timestamps; resolved by implementing collision-proof UUID suffixes.

Accomplishments that we're proud of

  • True Multi-Device Proactivity: Zero manual intervention required for complex multi-step cross-device workflows.
  • Full Compliance with Alexa+ MCP 2025-11-25: Clean separation of tools, resources, and live UI cards.
  • 10% Judging Bonus: Documented 5 comprehensive developer friction logs with exact reproduction steps in FRICTION_LOGS.md.
  • 100% Open Source: Fully documented, runnable in under 60 seconds with python run.py, under the permissive MIT License.

What we learned

We gained deep first-hand experience with Amazon's new Model Context Protocol specifications, AWS Bedrock Converse tool-use orchestration, and designing ambient multi-modal interfaces that inform without overwhelming the user.

What's next for Aetheris Omni-Agent

  • Native integration with physical Echo Show 21 and Echo Hub hardware.
  • Real-world Matter and Zigbee mesh protocol bridges for smart deadbolts and Phillips Hue.
  • Enterprise multi-tenant deployment for corporate executive offices.

Built With

  • alexa
  • amazon-nova
  • anthropic-claude-3-5-sonnet
  • aws-bedrock
  • bee-wearable-ai
  • boto3
  • css3
  • fastapi
  • fire-tv
  • html5
  • javascript
  • json-rpc-2-0
  • mcp-spec-2025-11-25
  • model-context-protocol
  • python
  • ring-api
  • server-sent-events
  • sqlite
  • streamable-http
  • vega-os
  • web-speech-api
Share this project:

Updates

Submission history