Inspiration
I focused on answering a user' overall health questions based on the data that they provided. I chose to solve this because I felt like this is a pertinent problem where people have all sorts of medical issues that are caused by factors that they might think is irrelevant. There are plenty of examples of people using tools like ChatGPT to figure out their medical problems where their doctors were unable to and that's partially because their medical providers might not ask them questions such as ‘what have you eaten over the past few days’ or ‘how has the intensity of your workout changed’. By allowing the user to upload all this data to one place we make it easier to physicians or AI assistants to diagnose and solve issues.
Tools / ML models you have used
For Development I used claude code and cline (with gpt-5 and claude sonnet) for dev work after the initial website was built by vercel v0.
For Features on the site I used the ElevenLabs Conversational AI to create a voice assistant that is getting users data and has tools to execute client-side tasks (eg displaying data, navigating to other tabs). I self hosted the AI model using Modal and host some of the data on Modal.
What has worked well with these tools
For Development There were times when the LLMs were not helpful at all. For instance, in trying to get the foodlogging to work with supabase the LLM for some reason was struggling so I had to be harsher in my tone in order for it to work especially once it kept going in circles. It ended up finding a slightly different solution which worked well.
What was challenging?
Honestly, just balancing where I spend my efforts. I wish I had spent less time trying to get the form workout analysis to work and instead spent more time building more tools into the voice assistant (eg getting it to suggest medical tests I should get done).
Built With
- gpt-oss
- llm
- modal
- typescript
- vercel
Log in or sign up for Devpost to join the conversation.