Aurai

Aurai is your AI health companion that talks like a human, not a bot-giving real insights from your wearable data through natural, face-to-face conversations.

Comment

🩺 Aurai

🧠 Inspiration

It all started with a shared frustration. Two of our teammates regularly use Oura Rings to track their health — and while the app provided detailed metrics on sleep, readiness, and activity, the experience often felt overwhelming.

There was so much data — but very little clarity.

They found themselves staring at dashboards full of numbers, wondering:
“What does this really mean for me? And what should I actually do about it?”

That inspired a question:
What if we could just talk to our health data?
What if we could ask questions like, "Will this 5-Hour Energy mess up my sleep?" — and get real, contextual answers based on the wearables we already use?

With that vision, Aurai was born.

💡 What it does

Aurai is your AI-powered health companion that makes your wearable data actually useful. It connects with your Apple Health and Oura data, and lets you engage in natural, face-to-face conversations — not chatbot scripts or wall-of-text dashboards.

Hold up an energy drink and ask, “How will this affect my sleep?”
It knows your recent sleep trends, your typical caffeine sensitivity, and gives a smart, relevant answer like:

“That’s got 200mg of caffeine. Based on your past data, expect deep sleep to drop by 15–20%.”

It’s not just smart — it’s personal, visual, and easy to use.

🛠️ How we built it

We used a full-stack architecture combining:

  • Frontend: Next.js with React
  • State Management: Zustand for lightweight, global state control
  • Backend: Django (Python), exposing REST APIs for data handling and AI logic
  • Database: SQLite — simple and fast for rapid prototyping
  • Integrations:
    • Oura API for sleep and readiness metrics
    • Apple HealthKit (via local sync) for general health data
    • OpenAI API for conversational understanding and insight generation
    • Tavus for personalized video-based responses, making conversations feel human

The result is a seamless, human-like interaction layer on top of deep biometric data.

🚧 Challenges we ran into

We hit plenty of roadblocks along the way, including:

  • Pulling BEFORE Pushing
  • Oura OAuth flow — Token exchange and refresh mechanisms were unexpectedly complex and time-sensitive, causing several delays.
  • Apple Health sync — Unlike cloud APIs, HealthKit required local device syncing, adding complexity to data ingestion.
  • State management across API calls — Coordinating live API data and conversation logic made us rethink our frontend structure until Zustand saved the day.
  • Prompt tuning for OpenAI — We iterated multiple times to make sure responses were not just accurate, but friendly, clear, and based on your data.
  • Tavus integration — Creating smooth video responses from live conversations added complexity but made a huge impact on the user experience.

🏆 Accomplishments that we're proud of

  • Built a working end-to-end system in under 48 hours.
  • Seamlessly integrated real-time health data with conversational AI.
  • Created a visual, voice-enabled experience that feels more like FaceTiming a doctor than using an app.
  • Gave users the power to understand and act on their health — not just watch it.

📚 What we learned

  • How to integrate and coordinate multiple APIs (OpenAI, Oura, Apple Health, Tavus) into a unified health experience.
  • How to use Zustand to manage global app state across complex component trees.
  • The ins and outs of Django REST framework, OAuth authentication flows, and building secure, clean backend logic.
  • How small UX tweaks — like adding facial video, or clarifying AI responses — can drastically improve user trust and usability.

🚀 What's next for Aurai

We’re just getting started. Next steps include:

  • Integrating more wearables using platforms like TerraAPI (pending cost support).
  • Adding longitudinal visualizations to track health changes over time.
  • Personalizing conversations even further with chat memory and follow-up tracking.
  • Building out real-time recommendations and nudges to help users make better choices in the moment.
  • Recommending medicines/remedies to users based on experienced symptoms

Health data should feel human.
With Aurai, we’re building the doctor’s assistant of the future — one conversation at a time.

Built With

Share this project:

Updates