Cursor for doctors

Inspiration

Doctors often lose time searching for trustworthy recommendations buried in lengthy guidelines, slowing clinical decisions.

What it does

“Cursor for Doctors” acts like “Perplexity for doctors”—instantly answering clinical questions using official guidelines, but does not yet propose actionable items like lab orders.

How we built it

We embedded guideline texts into a vector database and built an AI agent that reasons through clinical queries step-by-step to surface concise, guideline-backed answers.

Challenges we ran into

• Extracting clear, context-rich information from complex guideline documents
  
• Making AI outputs both trustworthy and easy to understand
  
• Limiting the system to informative suggestions (not direct clinical actions) for now

Accomplishments that we're proud of

• Real-time, relevant answers sourced directly from official guidelines
  
• Multi-step reasoning that outperforms simple keyword search
  
• Positive feedback on usability and clarity

What we learned

• Clinicians value speed and reliability above all
  
• The right answer often requires several layers of focused querying
  
• Actionable support (like lab order suggestions) is in high demand

What's next for Cursor for Doctors

• Enable automatic, guideline-driven action proposals (e.g., lab and imaging orders)
  
• Broaden guideline coverage and specialties
  
• Integrate directly with clinical workflows and EHRs
  
• Add source citation for full transparency

Built With

• n8n
• openai
• python
• qdrant
• streamlit