Introduction
NightShift MD is a focused triage and workflow assistant designed to eliminate chaos in night-shift medical clinics, where reduced staffing makes rapid, safe decision-making essential. The system allows patients to submit symptoms, automatically assigns a preliminary triage level, and provides doctors with a clear, real-time dashboard to manage incoming cases.
Developed using Next.js, Supabase, and Kiro, this project demonstrates how AI-assisted, spec-driven development can accelerate the creation of safe and structured tools for real clinical environments.
Live Demo: https://nightshift-psi.vercel.app/
Problem Overview
Night-shift medical facilities face several recurring operational challenges:
- Patients often wait without any initial assessment
- Staff receive unstructured or incomplete symptom descriptions
- No standardized triage workflow
- Difficulty tracking the status of multiple cases simultaneously
- Increased chances of miscommunication due to minimal staff
These issues contribute to delays and inconsistencies in patient handling, which can critically impact patient outcomes. Despite being widespread, many small clinics continue to rely on fully manual processes.
NightShift MD addresses these gaps with a clear and predictable system.
Solution Summary
NightShift MD provides a unified workflow that improves clarity for both patients and clinicians.
Patient Intake
A clean, mobile-first interface allows patients to quickly provide:
- Name
- Contact information
- Gender
- Reported symptoms
The form prioritizes speed and simplicity during high-pressure situations.
Automated Triage
The system analyzes symptom descriptions and generates a preliminary risk category:
- Low Risk
- Medium Risk
- High Risk
This ensures that clinicians immediately understand urgency levels.
Doctor Dashboard
Clinicians have access to a structured dashboard displaying:
- New and active cases
- Reported symptoms
- Triage level
- Patient contact details
- Case status
Doctors can update each case as:
- Contacted
- Treated
- Follow-up Required
- Emergency
The entire workflow is optimized for clarity and efficiency during night duties.
Live Application
Access the Live Project:
https://nightshift-psi.vercel.app/
The interface is responsive, with a patient-friendly mobile layout and a more comprehensive desktop dashboard for clinicians.
How Kiro Accelerated Development
This section outlines how Kiro contributed to rapid, structured, and safe development across the entire project lifecycle.Kiro wasn't just a code generator—it served as a full-stack development partner, providing both speed (Vibe Coding) and safety (Steering Documents) across five key areas:
A. Vibe Coding for Fast Implementation
Vibe coding enabled rapid generation of:
- Next.js pages and layouts
- API routes
- Supabase insert/update logic
- Form validation
- Reusable dashboard components
- Status update handlers
Requests such as “Generate a Next.js API route using this data schema and validate the fields” resulted in consistent, production-ready code.
This significantly reduced time spent on boilerplate and repetitive patterns.
B. Spec-Driven Development for Consistency
Given the sensitivity of medical workflows, accuracy and consistency were critical.
A clear project specification was defined, covering:
- Data models
- User flows
- Triage rules
- Page behaviors
- Design constraints
- Safety considerations
Kiro was instructed to generate code strictly aligned with the specification, ensuring predictable and uniform results across the entire application.
Vibe coding handled rapid iteration; spec-driven prompts handled precision.
C. Steering Documents for Safety Assurance
To maintain safety boundaries, a dedicated steering document guided all AI-generated outputs.
It enforced rules such as:
- No diagnosis or prescription suggestions
- Only triage classification
- Mandatory disclaimers where needed
- Neutral and calm language for patient-facing content
- Strict formatting guidelines
This produced controlled, reliable outputs aligned with medical communication standards.
D. Agent Hooks for Workflow Automation
Agent hooks helped automate repetitive development tasks, including:
- Generating structured summaries from raw symptoms
- Formatting case entries for the doctor dashboard
- Creating Supabase CRUD handlers
- Producing consistent schema-driven utility functions
Typical automated tasks included:
- Creating triage objects from patient symptom text
- Generating type-safe API handlers
- Converting raw input fields into validated objects
This dramatically reduced manual coding time and improved consistency.
E. MCP Integration for Database & Schema Management
MCP tools extended Kiro beyond chat-based assistance and enabled direct interaction with project resources.
They supported:
- Schema validation
- Database structure updates
- Test data cleanup
- Automated scaffolding for related components
This made Kiro function as a full-stack partner throughout development—not just a conversational model.
System Architecture
NightShift MD follows a clean and maintainable architecture:
Next.js (Frontend + API Routes)
│
Supabase (Database, Auth, Data Storage)
│
Kiro (Development Engine)
• Spec-driven code generation
• Automated triage logic creation
• Steering for safety
• Agent hooks for automation
• MCP for schema and data operations
This structure allows the system to remain lightweight, efficient, and scalable.
Impact and Future Expansion
NightShift MD can be adopted by small clinics, telemedicine platforms, and rural health centers to streamline nighttime patient processing.
- Future improvements may include:
- Multi-doctor access and collaborative dashboards
- Advanced analytics for patient trends
- Multi-language triage support
- Expanded symptom analysis
- Integration with hospital systems
- Triage escalation workflows
The current architecture supports incremental growth without major changes.
Conclusion
NightShift MD illustrates how AI-assisted development can produce a practical, real-world healthcare tool with speed and precision. Kiro contributed throughout the entire build process—from initial specifications to UI development, API logic, and safety enforcement.
This project demonstrates that combining structured development practices with AI tooling can significantly accelerate the delivery of reliable applications, especially in domains where clarity and safety are paramount. Explore the live demo today to see AI-assisted safety in action.