Inspiration

As students, we're so used to technology making our everyday lives easier that we sometimes forget how difficult certain tasks can be for other people.

We started thinking about elderly people, especially those who live alone, and how overwhelming it must be to manage multiple medications without always having someone around to help.

Remembering which tablets to take, at what time, how many are left, and when to collect another prescription might seem like small things individually. But having to keep track of all of this every single day must be exhausting, especially for someone who isn't very comfortable using technology.

We couldn't imagine how stressful it must feel to worry about forgetting something important, particularly when it comes to your health.

That's what inspired us to create medAssist AI.

We wanted to build something that could make life a little easier for elderly people. Not another complicated app with loads of buttons and confusing features, but something simple, welcoming, and easy to understand.

Our goal was to help people feel more organised and independent, while giving them one less thing to worry about in their day.

We know technology can never replace the support of family members, carers, doctors, or pharmacists, but we hope it can offer a little extra help, especially for people who don't always have someone nearby.

What it does

medAssist AI is an AI-powered medication management website designed to make keeping track of prescriptions simpler and less stressful.

One of our main features is the AI assistant. Instead of having to fill in a long form, users can explain their medication in everyday language.

For example, they could type:

"I collected 60 tablets today, and I take two every morning."

The AI helps turn this into a medication entry, which the user can check and confirm before saving.

We also built features that allow users to:

Record their medications, quantities, collection dates, and prescribed schedules.
View medication schedules and important dates on a calendar.
See estimates of when their medication supply might run out.
Receive in-app reminders when their estimated supply is getting low.
Archive medications while keeping a record of their history.
Keep important emergency contacts, allergies, and notes in one place.
Adjust the website's appearance using larger text, different themes, and high-contrast settings.
Create personal accounts to manage their own medication records.

We tried to make the whole experience feel as straightforward as possible, especially for someone who might not use websites very often.

The AI is there to help users enter information, not to give medical advice or make decisions for them. Users always have the chance to review their details before saving.

How we built it

We built medAssist AI as a full-stack web application using several different technologies.

For the frontend, we used React, JavaScript, and Vite to create the website and its interactive features. We focused on making the design clean, readable, and accessible, with clear navigation and options to adjust text size and contrast.

For the backend, we used Python and FastAPI to handle medication records, calculate estimated supply dates, manage reminders, and connect the website to our AI model.

We integrated NVIDIA Nemotron through Nebius AI to power our conversational assistant. This allows users to describe their medication naturally and have the AI help organise the information into a structured form.

We used Supabase for user authentication and database storage, with Row Level Security policies designed to ensure users can only access records belonging to their own accounts.

We deployed our frontend using Netlify and our backend using Render. We also used Git and GitHub to manage our code and collaborate as a team.

Challenges we ran into

One of our biggest challenges was getting the frontend, backend, AI, and database to work together. Sometimes everything looked fine on the website, but the actual functionality wasn't working as expected.

A particularly frustrating issue was getting medication records to save correctly.

During development, we discovered that our original demo setup wasn't properly separating medication records between users. Since we were building something involving sensitive information, we knew we needed to fix this.

We worked on adding individual user authentication and making sure medication records were associated with the correct accounts.

However, after connecting everything to Supabase, we kept getting errors whenever we tried to save a medication.

We spent ages trying to figure out whether the problem was in our frontend, backend, or database.

Eventually, by checking our backend logs, we discovered that Supabase's Row Level Security settings were blocking the requests because the necessary policies hadn't been created.

It was stressful, but it taught us how important it is to understand the systems behind an application rather than just making the interface look good.

Another challenge was making sure our AI was actually useful. We didn't want to add AI just for the sake of it. We wanted it to solve a real problem by making medication entry easier and less intimidating.

We also had to think about accessibility. As students who use technology all the time, it was easy to forget that things we find obvious might not be obvious to an older person.

This made us think more carefully about how we designed our buttons, forms, text, and navigation.

Accomplishments that we're proud of

We're really proud that we managed to turn our idea into a working web application.

At the beginning, medAssist AI was just an idea based on wanting to make everyday life easier for elderly people. Seeing it become an actual website with an AI assistant, medication calendar, and tracking features has been incredibly rewarding.

We're especially proud of integrating NVIDIA Nemotron into something with a practical purpose. Rather than creating a general chatbot, we built an assistant that helps users complete a specific task.

We're also proud of the accessibility features we included. We wanted the website to feel welcoming and easy to use rather than overwhelming.

Another accomplishment was working through the technical problems we encountered. There were definitely moments when we felt stuck, especially with authentication, database permissions, and deployment, but we kept trying until we understood what was going wrong.

Most importantly, we're proud that our project focuses on helping people.

Even if medAssist AI could make medication management a little less stressful for one elderly person living alone, that would mean a lot to us.

What we learned

We learned so much during this hackathon, especially because none of us had ever built an AI agent before!

We learned how to use technologies and platforms we'd never worked with before, e.g Render. We also gained experience working with APIs, integrating NVIDIA Nemotron into our application, managing databases, and deploying a full-stack project.

One of the biggest learning experiences was definitely debugging. There were so many times when something wasn't working, and we had absolutely no idea why. We learned how to read error messages, check backend logs, and gradually figure out what was going wrong. It was frustrating at times, but finally getting something to work after spending ages trying to fix it was such a rewarding feeling.

We also learned a lot about teamwork. Since we were all learning new things while building the project, we had to communicate, help each other out, and figure things out together.

Beyond the technical side, this hackathon also made us think more about the people we're designing technology for. We realised that creating something useful isn't just about writing code or adding impressive AI features. It's about making something that people can actually understand, feel comfortable using, and benefit from.

What's next for medAssist AI

Although we're proud of what we've built, we know there's still a lot we could improve.

One of the main things we'd love to add is real notifications for medication times and prescription refills. At the moment, our app provides a calendar and in-app reminders, but having optional notifications could make it even more useful.

We'd also like to improve support for people who have more complicated medication schedules, add more personalisation options, and continue improving accessibility.

Most importantly, we'd love to test medAssist AI with actual elderly users.

We can make assumptions about what might be helpful, but hearing directly from the people we're designing for would help us understand what works, what doesn't, and what needs to change.

We would also want to carry out further security, privacy, and reliability testing before considering the application suitable for real medical information.

Our hope is that medAssist AI could eventually become something that genuinely supports older people in their everyday lives, particularly those who live alone or don't always have someone around to help.