DocuEase

Inspiration

We were inspired by the frustration clinicians face every day with existing EHRs: too many clicks, redundant data entry, and endless alerts. Studies show providers spend 60% of their time documenting instead of caring for patients. We wanted to give that time back.

What it does

DocuEase is an AI-powered EMR co-pilot. It automatically summarizes patient timelines, drafts progress notes, proposes and schedules follow-ups, and flags critical labs or overdue items. Every action requires clinician confirmation and is fully auditable for compliance.

How we built it

• Frontend: Next.js with TypeScript and Tailwind for a clean, responsive UI
  
• Backend: Supabase with Postgres for database and auth
  
• AI Layer: Mastra for orchestrating agent flows, Cedar-OS for safe action execution with guardrails
  
• RAG: pgvector embeddings for context-aware retrieval of patient data
  

Challenges we ran into

• Implementing mastra model was a big challenge
• Designing a UI that reduces clicks while remaining familiar to clinicians
  
• Implementing agentic ai was a a big chanllenge as .

Accomplishments that we're proud of

• Built a working EMR demo that turns multi-step tasks into one-click workflows
  
• Implemented safety-first AI with confirmation and audit trails
  
• Reduced documentation burden and follow-up scheduling time in our demo flows
  

What we learned

• Simplicity matters more than adding features—one less click can mean a lot
  
• AI must be transparent and auditable to gain clinician trust
  
• How to embed retrieval-augmented AI safely in a clinical workflow
  

What's next for DocuEase

We plan to expand to multi-role support (clinician, admin), integrate real clinical guidelines into the RAG layer, and test with actual providers.

Built With

• cedar
• javascript
• mastra
• postgresql
• typescript