Luna AI

Luna is Loop Health’s AI health advisor — an intelligent conversational assistant that provides evidence-based guidance on peptide therapy, lab interpretation, health optimization, and protocol management.

Capabilities

Lab Analysis — Parse and interpret uploaded lab report PDFs with biomarker extraction and reference range analysis
Contraindication Checking — Verify peptide safety against patient conditions, medications, and biomarkers
Protocol Tracking — Log adherence check-ins and calculate streaks
Wearable Insights — Analyze sleep, recovery, HRV, glucose, and strain data with trend detection
Patient Context — Access full patient health profiles for personalized advice
Drug Interaction Checking — Identify potential medication interactions
Check-In Analysis — Review weekly check-in data, celebrate progress, troubleshoot issues

Architecture


User Message
      │
      ▼
┌──────────────────┐
│ Emergency        │  Detect crisis keywords
│ Detection        │  Return 911 guidance if needed
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│ Patient Context  │  Load profile, conditions, meds, labs
│ Assembly         │  Check for emergency flags
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│ AI Model         │  Anthropic Claude (default)
│ + 13 Tools       │  Up to 5 tool calls per response
│                  │  Temperature: 0.7
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│ Post-Response    │  Compliance guardrails
│ Guardrails       │  FDA, PHI, business, safety
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│ Logging          │  Clinical interactions logged
│ & Storage        │  Conversation history stored
└──────────────────┘

AI Provider

Luna uses Anthropic Claude as the default AI model, with OpenAI GPT-4 as a fallback. The provider is configured in the chat route:

Model: Claude Sonnet (configurable)
Temperature: 0.7
Max Steps: 5 (tool calls per response)
Streaming: Yes (Server-Sent Events)

Tools

Luna has 13 specialized tools across 7 categories:

Category	Tools
Health Data	`checkContraindication`, `getBiomarkerRange`, `getRequiredPanels`, `searchBiomarkers`
Patient Graph	`getPatientContext`, `checkDrugInteractions`, `getLabResults`, `getActiveProtocols`
Protocol	`trackProtocol`
Protocol Analysis	`analyzeProtocol`
Lab Analysis	`analyzeLabReport`
Wearable Insights	`getWearableInsights`
Check-In	`analyzeCheckIn`

See the Tools Reference for complete documentation of each tool.

Safety Features

Guardrails — Three-layer post-response validation (compliance, business, safety)
Emergency Handling — Crisis detection with 911 guidance
Medical Disclaimers — Always included in clinical responses
Contraindication Checks — Automatic safety validation before recommendations

Quick Links

Tools Reference — All 13 tools with parameters and examples
Guardrails — Compliance, business, and safety layers
System Prompt — Luna’s persona and behavior configuration
Emergency Handling — Crisis detection and response
Conversation History — Cross-channel conversation storage
Compliance Logging — Clinical interaction and escalation logging