LifeOS — AI Life Operating System

Your AI life operating system on Alexa+. Orchestrate tasks, focus, health & productivity through a single conversational MCP interface powered by AWS Bedrock.

Comment

Inspiration

We've all been there — seventeen browser tabs open, a calendar that looks like a game of Tetris gone wrong, and the nagging feeling that you're somehow busier than ever but getting less done. The promise of conversational AI assistants has always been that they'd help with exactly this kind of cognitive overload.

However, productivity tools remain deeply siloed.

Your calendar doesn't communicate with your task backlog. Your to-do list doesn't understand your energy cycle. Alexa can tell you the weather, but she couldn't look at your upcoming week and warn you that Thursday afternoon is going to derail you unless you protect your deep-work window.

When Amazon announced the Model Context Protocol

(MCP) integration for Alexa+, we realized this was the turning point. What if Alexa+ wasn't just a voice assistant, but an active Life Operating System — an ambient intelligence that understands your daily trajectory, context, and mental bandwidth, orchestrating your life through fluid dialogue?

That vision is **LifeOS**.

What it does

LifeOS is an enterprise-grade, self-hosted MCP server (spec 2025-11-25, Streamable HTTP) that equips Alexa+ with 7 modular life management superpowers:

- **`get_my_day`**: Synthesizes an ambient morning

briefing combining prioritized tasks, schedule friction points, and an AWS Bedrock focus mandate tailored to your energy levels. - manage_tasks: Complete natural language task operations (create, complete, reschedule, query) with automated multi-factor cognitive load prioritization. - analyze_productivity: Employs AWS Bedrock to identify productivity bottlenecks, velocity shifts, and actionable routine adjustments over longitudinal periods. - draft_message: Drafts high-fidelity correspondence across Email, Slack, and SMS, adapting tone dynamically between executive, assertive, casual, and diplomatic. - life_coach: Provides structured problem- solving sessions using strategic coaching frameworks (Socratic, Directive, Strategic Decomposition). - focus_mode: Launches adaptive Pomodoro sessions complete with task-specific mental framing, distraction defense tactics, and scheduled breaks. - daily_journal: Facilitates voice journaling with sentiment classification, longitudinal mood tracking, and recurring emotional theme identification.

How we built it

LifeOS is engineered with a modular, 4-tier architecture:

1. **MCP Server Core (spec 2025-11-25, Streamable

HTTP)** - Developed in TypeScript/Node.js utilizing the official @modelcontextprotocol/sdk and Express. - Built to meet Alexa+ Streamable HTTP transport specifications. - Full OAuth 2.1 compliance with PKCE (S256 mandatory) and RFC 9728 Protected Resource Metadata at /.well-known/oauth-protected-resource alongside RFC 8414 Authorization Server Metadata at /.well- known/oauth-authorization-server. - Returns clean 401 Unauthorized responses without WWW-Authenticate headers as mandated by the Alexa+ integration spec.

2. **AWS Bedrock Intelligence Runtime (Amazon Nova

Pro)** - Integrated via @aws-sdk/client-bedrock- runtime calling amazon.nova-pro-v1:0. - Abstracted through an enterprise client layer in server/src/lib/bedrock.ts handling retry budgets, prompt templating, structured JSON responses, and zero-downtime fallback to rule-based logic when operating offline.

3. **Interactive Multi-Modal Web Simulator**
   - Built with React, Vite, Tailwind CSS, and

Lucide icons. - Directly executes JSON-RPC 2.0 tool calls against the live MCP server, providing judges with a live, functional testing environment.

4. **Production Cloud Deployment**
   - Packaged with a multi-stage Alpine Docker

container running non-root security privileges (USER lifeos). - Deployed on edge infrastructure via Fly.io with persistent volumes for task/journal state.

Challenges we ran into

  • Strict Alexa+ OAuth 2.1 & RFC 9728 Compliance: Alexa+'s integration security strictly requires RFC 9728 Protected Resource Metadata, mandatory PKCE S256, and suppression of the WWW- Authenticate header on unauthenticated 401 responses. Aligning the metadata and token exchange routes across the protocol bridge required precise implementation.
    • Conversational Protocol Tool UX: Designing MCP tools for voice-first interactions requires strict input validation schemas (zod) that gracefully handle incomplete user utterances while returning concise text for speech synthesis and structured payloads for multimodal screens.
    • Inference Latency Optimization: Synchronous LLM calls in tool responses risk voice timeouts. We streamlined system prompts, applied structured output schemas, and cached contextual briefing metadata to maintain sub-second response times. ## Accomplishments that we're proud of
  • 100% Spec-Compliant Remote MCP Server: Fully compliant with the latest MCP 2025-11-25 Streamable HTTP protocol.
    • Deep AWS Bedrock Nova Pro Integration: Seamless AI orchestration powering structured life coaching, sentiment trends, and task prioritization.
    • End-to-End Live Deployment: Accessible 24/7 with a public endpoint, live health checks, and a functional browser simulator.
    • Robust Security: Enforced OAuth 2.1 PKCE (S256), scoped JWT validation, and non-root execution containers.
    • 15 High-Resolution Architectural Slides & Video: Created a complete 105-second demo video and a 15-slide asset gallery directly documenting every subsystem. ## What we learned
  • The Model Context Protocol (MCP) Streamable HTTP standard is poised to become the universal communication bus between LLM reasoning engines and device capabilities.
    • Designing voice-first agentic tools requires a paradigm shift: tool responses must be optimized for audio comprehension first, with visual cards providing progressive enhancement. ## What's next for LifeOS — AI Life Operating System
  • Wearable Biometric Fusion (Bee Integration): Ingesting heart-rate variability and physical recovery scores to automatically adjust task schedules.
    • Fire TV Focus Companion: An ambient 10-foot UI dashboard displaying current Pomodoro countdowns, daily priority cards, and gentle ambient soundscapes.
    • Calendar Bi-Directional Sync: Native Google Calendar and Microsoft 365 OAuth integrations.

Built With

Share this project:

Updates

Submission history