HomeRelay AI

• 

  Overview of the household workspace, bringing open commitments, pending confirmations, and today's timeline into one place.

• 

  Track household commitments by owner, due time, and status, with explicit confirmation before tasks are marked complete.

• 

  A self-hosted MCP server using Streamable HTTP, connected to the console with seven discoverable tools.

• 

  A successful MCP tool call approves four proposed commitments and adds Snacks and Drinks to the demo shopping list.

Inspiration

Everyday household plans often get lost between conversations, shopping lists, and reminders. People forget tasks, assume someone else has handled them, and struggle to keep track of what is actually completed. We wanted to explore how an AI agent could solve the problem of follow-through, not just answer questions.

What it does

HomeRelay AI is a conversational household coordination assistant. Users describe what they need in plain language, and the system breaks the request into manageable actions, such as errands, shopping items, and reminders.

Its signature feature is the Promise Ledger, a central dashboard that tracks each task's owner, deadline, and status. Tasks remain pending until completion is confirmed instead of being marked complete simply because the AI suggested an action.

How we built it

We are developing HomeRelay AI as a web-based prototype with a conversational interface and a visual task-management dashboard. Our implementation plan includes structured task creation, multi-step request handling, persistent task storage, and clear task-status updates.

The prototype will demonstrate these workflows using simulated household actions rather than claiming to control real smart-home devices. We will document the actual technologies, tools, and implementation details used in the completed version.

Challenges we ran into

One important challenge is translating everyday language into clear, actionable tasks without making incorrect assumptions. Another is keeping planned, pending, and completed actions separate so that users can trust the status displayed by the assistant.

Our design focuses on transparent task states, clear confirmations, and asking for clarification when a request is ambiguous.

Accomplishments that we're proud of

We identified a specific problem to solve and designed the Promise Ledger as the central feature of our solution. Rather than focusing only on conversational AI, HomeRelay AI emphasizes accountability, coordination, and visibility into the progress of everyday tasks.

Our goal is to demonstrate these ideas through a simple, polished, and useful interface.

What we learned

Planning this project highlighted the difference between generating an answer and completing a useful workflow. An effective assistant needs to manage context, organize actions, handle uncertainty, and communicate what has or has not actually happened.

We are using these principles to guide the implementation and testing of HomeRelay AI.

What's next for HomeRelay AI

Our next steps are to complete the interactive prototype, test it with realistic household requests, improve the dashboard experience, and demonstrate the full journey from a user request to confirmed task completion.

Future improvements could include real calendar integrations, notifications, shared household accounts, and supported smart-home integrations.

Built With

• alexa+
• amazon
• browser
• css
• local
• react
• simulated
• storage
• typescript
• vite