A caregiver describes the moment; CalmGuide returns a structured four-part response (Right Now, Why, What NOT to Do, When to Escalate), personalized by the patient’s stored profile and grounded in retrieved guidance from trusted sources. In 11 languages.
SAFETY GATERETRIEVEGUIDE
the safety path does not depend on an LLM
Each of these exists in the codebase. We name the mechanism so you can hold us to it.
Hybrid retrieval (vector similarity plus keyword match plus reranking) over guidance from the Alzheimer’s Association, Mayo Clinic, CDC, HelpGuide, and Family Caregiver Alliance. Retrieved context is injected into every coaching response.
Life-threatening and self-harm patterns are matched before any model call and short-circuit to structured emergency guidance. The safety path does not depend on an LLM behaving.
Patient names never leave the browser. The server works with SHA-256-hashed access codes and AES-256-GCM encryption at rest.
Episode-cycle detection and nightly analysis turn logged moments into insights and a "Tonight’s Outlook" risk read for the caregiver.
Full i18n including RTL scripts, with voice input on mobile.
Anonymized notes on what has helped similar patient profiles feed back into guidance.
iOS and Android apps built with Expo alongside the web app, sharing one backend.
A B2B surface for facilities: dashboards, incident tracking, resident and staff views, and a care-journey module from first noticing through bereavement.
Where it standsBeta with a deployed production API; the mobile app is built and app-store submission is in progress. CalmGuide supports caregivers. It is not medical advice.
AI phone receptionist that answers calls and books the appointment.
Open-source AI support that resolves customer questions and escalates the hard ones.
MCP gateway that gates every agent tool call.