CarrySafe

Supervisor review
Integrity center
Shift handovers
Shift handovers overview
Operations
Assistant response
Assistant home

INSPIRATION

In factories, work doesn't stop when one shift ends. Another team takes over, and they need to know what happened during the previous shift.

We thought about a situation where a worker finds a leaking pipe. They report it and place a bucket underneath as a temporary solution. But the actual repair is still pending. When the next shift comes in, how will those workers know about the leak? What if nobody checks whether it has been repaired?

This made us think about how small communication gaps can become serious safety problems.

We wanted to build something that could keep track of these unfinished issues and make the information available to the next shift. That's how we came up with CarrySafe.

WHAT IT DOES

CarrySafe is a web application that helps factories manage incidents and transfer important information between shifts.

Workers can report problems, record temporary safety measures and track whether the issue has been resolved.

We built an Operations page where users can see ongoing incidents, pending actions and overdue reviews.

There is also a Shift Handovers section that helps the morning, evening and night shift teams understand what still needs attention.

One of the main parts of our project is the AI assistant. Instead of searching through different pages, a user can ask the assistant to brief them about the handover.

The assistant uses MCP tools to retrieve relevant information from the application.

We also added checks for missing responsibilities, overdue reviews and incomplete handovers. Important decisions, such as closing an incident, still require human approval.

HOW WE BUILT IT

We used React and TypeScript for the frontend and FastAPI with Python for the backend.

We wanted the website to be simple enough for workers to use without spending too much time learning it.

For the AI functionality, we worked with Model Context Protocol (MCP). We connected MCP tools to our backend so the assistant could retrieve incident details and handover records instead of relying only on generated responses.

We also created different sections for incident tracking, handovers, supervisor reviews and audit history.

For the Alexa+ track, we focused on making these workplace operations accessible through conversational requests using MCP. Our current prototype demonstrates this through a web assistant rather than a direct Echo-device integration.

CHALLENGES WE FACED

Connecting the assistant to real application data was one of the more difficult parts.

It wasn't enough to make a chatbot answer questions. We needed it to retrieve the correct incidents and understand which information belonged to a particular shift.

We also had to think carefully about permissions. For example, an AI assistant shouldn't be able to close a safety incident just because somebody asks it to.

We worked on keeping those decisions within the application's approval process.

ACCOMPLISHMENTS WE'ER PROUD OF

We managed to bring incident reporting, temporary safety measures, shift handovers and an MCP-powered assistant into one application.

Our favorite part is the handover briefing. A user can request a summary from the assistant and see the tool activity used to retrieve the information.

We also built a demonstration workflow that follows an issue from its initial report through the next shift.

WHAT WE LEARNED

Before this project, we mainly thought about AI assistants as tools that answer questions.

While building CarrySafe, we learned how an assistant can connect with application functions using MCP and help users access information already stored in a system.

We also understood why safety-related applications need proper records, permissions and human verification.

WHATS NEXT

We want to improve the voice-based experience and explore integrating CarrySafe with Alexa+ through supported developer capabilities.

We also plan to improve notifications, add more realistic workplace scenarios and make the interface easier for workers to use.

Our goal is to make sure that important safety information doesn't get lost when one shift hands over work to another.