Care Circle

The Problem

When a family is caring for a loved one, coordination can become a job of its own.

One person handles an appointment. Another picks up supplies. Someone else takes over later, but important information gets lost between conversations and messages. Families can end up asking the same questions: What has been done? What still needs attention? What does the next caregiver need to know?

I wanted to explore how AI could make those everyday handoffs clearer, more organized, and less stressful.

Meet Care Circle

Care Circle is an AI-assisted care coordination app designed to help families stay on the same page.

It brings the essential parts of coordinating care into one workflow:

• Capture care updates: Turn everyday language into proposed tasks that can be reviewed before saving.
• Plan the day: See care tasks and understand what still needs attention.
• Track progress: Mark tasks as completed or reopen them when plans change.
• Create caregiver handoffs: Organize completed tasks, outstanding items, upcoming responsibilities, and information the next caregiver needs.

The goal is simple: make it easier for one caregiver to hand over responsibility without leaving the next person guessing.

Care Circle supports coordination, not medical decision-making. It is not intended to diagnose conditions or recommend medication dosages.

How I Built It

I approached Care Circle as a practical AI product rather than a chatbot that simply answers questions.

The project uses TypeScript and Node.js, with an Express backend and a web interface. I built around the Model Context Protocol (MCP) to explore how care workflows can be exposed as structured tools that an AI-enabled experience can use.

The intended workflow connects natural-language updates to reviewable tasks, persistent care records, daily planning, task-status changes, and handoff generation. The project also explores a web-based experience for demonstrating these interactions.

My focus was on making the underlying actions explicit and reviewable. A care update should not silently become a confirmed responsibility, and a handoff should reflect recorded information rather than invented details.

Challenges and What I Learned

One of the biggest challenges was moving from a working local development setup toward a deployable application with persistent data. Serverless deployment introduces different constraints from running an application locally, especially when database connections and file storage are involved.

I also had to think carefully about trust. In a care context, a polished interface is not enough. The system needs to distinguish between completed and outstanding tasks, make caregiver responsibilities clear, and avoid presenting assumptions as facts.

Building Care Circle strengthened my understanding of AI-assisted development, backend architecture, MCP, deployment, and the importance of designing workflows around real human needs.

I used AI as a development partner to help reason through architecture and implementation, while keeping the product problem and its intended user experience at the center.

What's Next

Care Circle is an early-stage project. The next steps are to improve the caregiver handoff experience, refine the interface, test the workflows more thoroughly, and make the application more reliable for families coordinating care together.

I want to keep building toward a tool that helps families spend less time reconstructing what happened and more time supporting the person who needs care.

Care Circle: clearer plans, better handoffs, and fewer things falling through the cracks.

Built With

• ai
• express.js
• libsql
• mcp
• node.js
• sqlite
• turso
• typescript
• vecrel
• zod