Care Ring

Framing the Problem

We aim to solve the problem of those people who are very busy and thus unable to take care of their older family members, even as simple as checking in with them everyday. This can be people who are working full-time, working multiple jobs or people with family who are living in different countries.

Idea Explanation

What is your idea? How does it fix the problem?

Our app CareRing provides an AI companion voice call service that is used to check up on the elderly. This service asks basic check in questions such as how was your day, what did you do and so on. It also has a function that can remind the elderly person to take their medication and also alerts the caregiver if the elderly person missed a call. CareRing also allows the caregiver to connect more with their elderly family members by implementing a feature where the user can schedule a call, a notification is sent when the time to call has arrived.

The first feature gives the elderly some companion who is able to check up regularly on them and remind them to take care of themselves in a way that the busy user may not be able to. Its alert system also allows the user to call the elderly person if anything goes wrong. The second feature allows the family to connect more by allowing both sides to schedule a call easily.

Implementation

How do all the pieces fit together? Does your frontend make requests to your backend? Where does your database fit in?

Our app first requires the user to choose whether they are a caregiver or an elderly person. By signing up to our app, the user also has to create a Twilio account and a Twilio phone number. Behind everything, We do all the front-end using HTML,CSS and JS. For the call itself we use Twillo to call the users and Gemini AI to give replies.

Challenges

Since we were introduced to many different things at once, like learning how to use Twilio, Google Gemini, ngrok, etc. We found it difficult to understand how to use all of them and integrate ethem into our program. Setting up ngrok for the first time was confusing. Since our server was running locally, Twilio needed a public URL to reach it, we also had problems deploying our website so it all ran locally. These challenges ultimately helped us better understand how real-world voice systems connect APIs, web servers and telephony infrastructure together.

Accomplishments

One of our biggest accomplishments was successfully building a system where an AI can call an elderly person through Twilio and have a real conversation with them. The AI was able to give very human responses and we believe it would be effective in real life.

Next Steps

What are the next steps for your project? How can you improve it?

We could implement our summary, (currently a prototype) where the AI gives a shortened version of the elderly's responses, so our caretaker could have a recap about their day. There should be a button to connect the elderly to the caretaker in case of an emergency, so we would also require the caretaker's phone number.

Tech Stack

Frontend:
We built the user interfaces using HTML, CSS and JavaScript.

Backend:
The backend is built with Node.js and Express, which manages the call flow, processes requests from the website and connects the different services used in the system.

AI & Conversation:
We used Google Gemini API to generate the conversational responses during the call.

Voice & Telephony:
Twilio Voice powers the phone calls between the AI system and the elderly user. Twilio handles the call connection and speech input, allowing the AI to respond in real time.

Local Development:
During development we used ngrok to expose our local server to the internet so Twilio could communicate with it while testing phone calls.