Local-first · Private alpha · Safety-reviewed AI

An AI-native personal health record, not another chatbox

Medora turns messy health information — glucose notes, CGM-style reports, lab snippets, medication text, symptoms, visit context — into a longitudinal, structured, source-grounded health record with timeline events, safety-reviewed doctor-prep summaries, and an auditable AI workflow history.

Glucose / metabolic first vertical
13+ record entities
Source-grounded insights
Local-first by default
WorkflowRunner AIWorkflow LLMProvider Zod validation SafetyGuard persistence
Every AI output is schema-validated and safety-reviewed before it ever touches your record.
The thesis

The durable center is the record — the model is just one worker in it

Most "AI health" tools render a model response straight into a UI. Medora inverts that: the longitudinal health record and the workflow layer are the product. The model is a replaceable provider behind a boundary, its output validated, safety-checked, and grounded back to its source before it becomes part of your history.

  • Structured extraction — free-text and reports become typed metrics, readings, meds, symptoms.
  • Longitudinal timeline — events accrete over time instead of vanishing into a chat scrollback.
  • Source-grounded — every insight links back to the document it came from.
  • Auditable — every agent run, safety review, and blocked output is inspectable.
Capabilities

Real health-record surfaces, not a dashboard clone

Starting with glucose, diabetes, and metabolic health — with an architecture deliberately broad enough to grow.

🧬

Structured extraction

Glucose notes, CGM-style reports, lab snippets and medication text are parsed into typed entities and Zod-validated before persistence.

📈

Longitudinal timeline

Readings, metrics, meds, symptoms and visits become durable timeline events you can trace across months — not a disposable thread.

💡

Source-grounded insights

Insights cite the exact source reference they were derived from, so nothing is a free-floating model assertion.

🩺

Doctor-prep summaries

Safety-reviewed visit summaries that organize what to bring up — explicitly not diagnosis, triage, or dosing.

🛡️

Medical safety guard

A rule-based guard blocks prohibited output classes; blocked outputs are logged and inspectable, with a safety-regression suite.

🔄

Provider abstraction

A deterministic mock provider for local dev; OpenAI behind the same LLMProvider boundary, enabled by env only.

/ profile
/ inbox
/ timeline
/ insights
/ doctor-summary
/ privacy-safety
/ onboarding
/ alpha-access
Safety · Privacy

Health data earns the highest bar — so the defaults are conservative

Medora is a local-first private-alpha MVP for a small, closely monitored test group. It is honest about what it is not.

  • Local-first by default; the OpenAI provider activates only via explicit environment configuration.
  • User-owned data with JSON export, deletion, and blocked-output inspection.
  • Every AI output is Zod-validated and passes a safety guard before persistence.
  • Validation suites cover ingestion, workflows, safety regression, E2E flow, and alpha readiness.

Medora explicitly does NOT claim

  • production security or HIPAA compliance
  • regulated medical-device readiness
  • emergency triage capability
  • diagnosis or prescription authority
  • insulin-dosing capability
Architecture

A workflow pipeline, not a prompt-to-UI shortcut

Next.js + a Prisma-backed model spanning profiles, documents, extractions, metrics, glucose readings, medications, symptoms, timeline events, insights, doctor summaries, source references, safety reviews, feedback, and agent runs.

🗃️

Durable data model

13+ Prisma entities make the record — not a model response — the system of record.

⚙️

Workflow runner

WorkflowRunner → AIWorkflow → LLMProvider → Zod → SafetyGuard → persistence, each stage independently testable.

📋

Documented end-to-end

ARCHITECTURE, DATA_MODEL, MEDICAL_SAFETY, STATE_MACHINES and an MVP scope — the design is written down, not implied.