Introduction

HeyGran! is an AI-powered elderly care communication system built at HuddleHive's WIT Hackathon #5. Our challenge was to bridge the gap between AI and human connection experience — and we found our answer in one of the most human problems there is: the quiet distance that grows between families and the elderly people they love, where good intentions and busy lives too often get in the way.

How To Test

1. Click 'Login' to log in (you don't need to input user details)

1. Go to 'Loved Ones' tab

2. Click on a User's profile

3. Enter YOUR phone number

4. Click 'Trigger AI Check-in'

5. Accept the call and chat!

6. tips: you can say you are lonely to test the 'connect with someone'' feature

The Problem

According to NHS, more than 2 million people in England over the age of 75 live alone, and more than a million older people say they go over a month without speaking to a friend, neighbour, or family member. For families, the guilt is real — but daily calls are hard to sustain alongside work and life. Health changes go unnoticed. Medication gets missed. Small signs of confusion or distress slip through the cracks until they become something serious.

The problem affects everyone: elderly people who feel isolated, and the families who love them but can't always be there.

Our Solution

HeyGran! places a friendly AI phone call to your elderly loved one every day. No app for them to download. No screen to navigate. Just a warm, familiar voice asking how they're doing, whether they've taken their medication, and what's on their mind. The elderly person can also call or text the number any time they need support.

After every interaction, HeyGran! generates a summary for the caregiver, flags anything that needs attention, and updates a simple dashboard — turning everyday conversations into a continuous picture of health, mood, and wellbeing.

Tech Stack

HeyGran! is built across four connected layers:

Frontend

• Next.js (React) hosted on Vercel

• Tailwind CSS for styling

Backend

• Python / FastAPI hosted on Render

• WebSocket server for real-time voice conversation handling

AI / LLM

• OpenAI GPT-4o-mini for conversational responses, risk classification, and context-aware wellness check-ins

Voice & Telephony

• Twilio Programmable Voice with ConversationRelay for real-time AI phone calls

• ElevenLabs text-to-speech (via Twilio ConversationRelay) for natural-sounding voice

• Twilio SMS for caregiver alerts

• Twilio Conference for connecting lonely users with companions

  The frontend sends requests to the backend to manage call schedules and retrieve history. Incoming calls and messages are processed by the backend and passed to ChatGPT, which generates responses and summaries before storing results in the database.

Challenges

The biggest challenge was scope. The idea had many possible directions, and narrowing it to a realistic MVP required hard decisions about what was essential versus aspirational. Connecting Twilio, AI prompting, scheduling, and summary generation into a coherent flow — without a traditional app on the elderly user's side — required rethinking the architecture from the user backwards. On top of that, the topic itself — elderly loneliness, health monitoring, cognitive decline — is inherently serious, and we worried the project could come across as worthy but dull. We spent real time figuring out how to make the experience something people actually look forward to.

One of our most ambitious technical challenges was making the social feature work. We wanted Hey Gran to do more than just call one person — we wanted it to connect people. Using Twilio, we built a feature where the AI initiates two separate calls and bridges them into the same conversation, so a grandparent and someone else who also feels lonely can actually talk to each other, prompted and facilitated by the AI. Getting the call orchestration, timing, and handoff working reliably was a real puzzle.

Accomplishments

We are proud of building something that feels genuinely useful and emotionally meaningful, not just technically interesting. Key accomplishments include successfully integrating Twilio Voice and SMS with an AI backend, designing a caregiver dashboard that surfaces the right information at the right time, and developing a prompt system that keeps conversations warm, safe, and personalised. We also shaped a clear escalation logic — so when something urgent is mentioned, the right person is notified immediately.

Next Steps

With more time, we would first focus on making the core product more robust — improving real-time call handling, strengthening the escalation system for urgent alerts, and adding proper privacy, authentication, and secure data storage.

But the longer-term vision goes much further. HeyGran! has the potential to evolve from a daily check-in into a full cognitive care and connection platform. Features we would build next:

• Always-available companion — the elderly person can call HeyGran! any time they want a chat, not just during scheduled check-ins, so there is always a friendly voice available whenever they feel lonely or just want to talk

• Early dementia detection — the AI tracks speech patterns, word choice, and response speed over time to identify early signs of cognitive decline, flagging changes for caregivers and medical professionals

• Cognitive coaching — short daily games woven into the conversation, such as puzzles, riddles, and guess-the-song, designed to keep the mind active without feeling clinical

• Life memoir — with the user's consent, the AI periodically prompts them to share a personal story or memory, gradually building a written memoir of their life for family to treasure

• Persistent memory — the AI remembers previous conversations and refers back to them naturally, so every call feels like catching up with a friend rather than starting from scratch

The goal is a trusted AI companion that doesn't just monitor — it connects, stimulates, and remembers. Something that makes life richer for the elderly, and gives families not just peace of mind, but a deeper window into the lives of the people they love.