GeneClaw

GeneClaw uses your genetic data to monitor symptoms through regular check-ins, ask targeted questions, and detect early risks and automatically connects you with the right specialist when needed.

Comment

Inspiration

The story of this project is inspired from a personal experience of mine. This summer my parents visited me in Vienna and I wanted to do something meaningful and useful for them. As I am a biologist myself, I decided to have their DNA tested by a genetic test company. After a month, the result was ready and I ended up with a wealth of information that I have no idea how to implement them in real life. It require a team of multi-disciplinary people to interpret the data. So why not build an agent to help us with this task?

What it does

The system is an agent-based Telegram bot that uses genetic risk data to guide users through symptom-based screening.

It asks adaptive follow-up questions:

  • Evaluates potential risk in real time

  • Recommends further action when necessary

  • Connects users with specialists if required

How we built it

Due to the sensitivity of genetic data and restricted access to the original platform, we developed a dummy website that replicates the relevant structure and content, focusing only on high-risk conditions.

This allowed safe data access and scraping without exposing credentials.

We then built:

  • A Telegram bot interface for user interaction

  • An agent loop for adaptive questioning

  • A state management system to track responses

Challenges we ran into

  • Handling sensitive genetic data while maintaining privacy

  • Sending scheduled messages to the user in the chat bot

Accomplishments that we're proud of

  • Transforming complex genetic data into an interactive system

  • Building a functional agent-based medical questioning flow

  • Implementing a privacy-conscious data pipeline

  • Bridging genetic risk with actionable user guidance

  • Making sense of complicated medical terms and turn it into real-life applicable suggestion

What we learned

  • The importance of combining genetics, AI, and user-centered design

  • How to convert static reports into dynamic interactions

  • The complexity of designing adaptive agents

What's next for GeneClaw

  • The agents should know which disease is more urgent and schedule follow-up questions based on that.
  • Some symptoms are shared between different diseases and a biology-aware agents must be able to connect the dots and consider all the possible scenarios.

Built With

  • elevenlabs
  • github
  • lovable
  • maritime
  • openclaw
  • telegram-api
Share this project:

Updates

Submission history