Case Study 1 · Luna AI Companion
Voice-first AI care, with a visual layer that actually shows up.
Real-time, low-latency conversational AI orchestrating patient records, wearable data, and retail inventory.
Luna blends empathetic voice conversations with dynamic visual UI. When data matters — vitals, product options, prescription details — the agent shows it instead of reading it. Built on sub-second WebRTC, multi-agent routing, and hardcoded safety guardrails that bypass the LLM in emergencies.
- SmallWebRTC
- Pipecat
- Gemini Flash
- Deepgram Flux
- Cartesia Sonic-3
- Google Agent Platform Vector Search
- GCP Compute Engine MIGs
- React + Capacitor
Live Conversation
Luna · Triage Agent
Latency · 740 ms
Sub-second WebRTC · Pipecat orchestrating Gemini Flash + Cartesia Sonic-3.
Operational Scale
Intercepts routine triage and admin tasks, freeing nurses and pharmacists for critical care.
Proactive Engagement
Bridges passive wearable data and active clinical conversation — drives adherence.
Regulatory First
Anti-hallucination prompts and hardcoded emergency interception nodes route severe symptoms to 911.
Accessibility
Voice-first lowers barriers for elderly and visually impaired users; visual UI handles complex data.
Use Cases
What it powers
Vulnerable Patient Care
Syncs EMR and wearable data (Whoop, Google Health) for symptom triage, medication reminders, and 7-day vital trend reviews.
Pharmacy Product Finder
In-store assistant with Symptom, Vision, and Nurse sub-agents. Guides shoppers to physical shelf locations.
Digital Health Product Finder
E-commerce adaptation pointing to a tailored vector store of online inventory.
Dentist Office Assistant
Front-desk automation handling dental queries and routing patients to appointment booking.
Student Enrollment Guide
Multilingual translator for parents navigating school enrollment and DOH immunization requirements.
Veterans & Accessibility Care
Proactive care coordination and accessible follow-ups tailored for veterans and patients with disabilities.