v1 stable · prerelease

The inference layer for human biology — as an API.

Three endpoints. Signed responses with full provenance. Hardware- agnostic ingest via Terra-shape adapters, Apple Shortcuts, and user-mediated export. Build biological inference into any app — consumer wellness, employer cohort dashboards, supplement attribution, clinical adjacency. The data stays at the layer; you ship the insight.

01Endpoints

Three endpoints. That's the whole API.

Continuum's surface area is intentionally small. Connect data, ask questions, read history. The complexity lives at the layer; your app ships the insight.

POST/v1/infer

Cross-modal biological reasoning

Submit a natural-language query against a user's connected sources. Continuum runs sensor fusion across vendors, weights signals by confidence + cross-vendor agreement, and returns a signed inference with full citation lineage. Your app shows the insight; never sees the raw data.

Request
POST api.continuum.health/v1/infer
Authorization: Bearer ${secret}

{
  "user": "usr_8f2a91",
  "query": "why am i tired today?",
  "sources": ["auto"],
  "context": {
    "surface": "consumer-app",
    "audience": "self"
  }
}
Responsesigned
200 OK · 720ms · signed

{
  "insight": "...",
  "shortAnswer": "Three signals from four sources converged...",
  "citations": [
    { "source": "dexcom", "weight": 0.32, "confidence": 0.98 },
    { "source": "oura",   "weight": 0.22, "confidence": 0.94 },
    { "source": "whoop",  "weight": 0.18, "agreement": 0.96 },
    { "source": "myfitnesspal", "weight": 0.18 },
    { "source": "apple-health", "weight": 0.10 }
  ],
  "qualityScore": 0.92,
  "signature": "0x4f1c..."
}
POST/v1/connect

Hardware-agnostic ingest

Connect a user's wearable, CGM, or wellness app via OAuth (Terra-shape adapter), Apple Shortcuts payload (server-side restriction proof), or direct export upload. Continuum normalizes vendor-specific schemas into 23 standardized fields and computes per-source confidence scoring on first ingest.

Request
POST api.continuum.health/v1/connect
Authorization: Bearer ${secret}

{
  "user": "usr_8f2a91",
  "method": "oauth",
  "vendor": "oura",
  "redirect": "https://yourapp.com/cb"
}
Responsesigned
200 OK

{
  "url": "https://api.continuum.health/oauth/oura/...",
  "alternates": {
    "shortcut": "https://...icloud.com/shortcuts/...",
    "export": "POST /v1/connect/export with file=..."
  }
}
GET/v1/timeline

Synthesized longitudinal signal

Pull a user's lifetime baseline — synthesized across every vendor they've ever connected. Apple Watch 2024 + Whoop 2025 + Oura 2026 collapse into one continuous record with cross-vendor agreement scoring per field. The persistent biological record that follows the user across hardware churn.

Request
GET api.continuum.health/v1/timeline
  ?user=usr_8f2a91
  &fields=hrv,sleep,glucose
  &range=90d
Authorization: Bearer ${secret}
Responsesigned
200 OK

{
  "user": "usr_8f2a91",
  "range": { "from": "2026-01-25", "to": "2026-04-24" },
  "fields": {
    "hrv":     { "median": 52, "iqr": [44, 58], "agreement": 0.96 },
    "sleep":   { "median": 78, "iqr": [70, 84], "agreement": 0.94 },
    "glucose": { "median": 96, "iqr": [88, 104], "agreement": 1.0 }
  },
  "vendors": [
    { "id": "oura",    "first": "2026-01-25", "records": 90 },
    { "id": "whoop",   "first": "2026-02-12", "records": 72 },
    { "id": "dexcom",  "first": "2026-03-04", "records": 51 },
    { "id": "apple-health", "first": "2024-08-01", "records": 998 }
  ]
}
02Foundations

What's under every endpoint.

Signed inference

Every response carries a signature over (insight + citations + timestamp). Audit trails for HIPAA, BIPA, and enterprise legal review come free.

Raw data never leaves

Inferences leave the layer. Raw biometric streams stay encrypted at rest. Your app sees the insight, your customer's GC sees the architecture.

Hardware-agnostic, by design

Apple, Oura, Whoop, Garmin, Dexcom, MyFitnessPal, Strava — plus 400+ via Terra-shape adapters. iOS Shortcut bypass for Apple lockdown. Export upload as the resilience path.

Sub-second median inference

Claude Sonnet 4.6 with prompt caching. P95 under 1.4s for typical queries. Built for real-time consumer surfaces and workflow integrations.

03Pricing

Free for builders. Usage-priced for platforms. Concentrated for enterprise.

Developer

Free

10K inferences / mo

  • ·Up to 1,000 connected users
  • ·All three endpoints
  • ·Signed responses
  • ·Community support

Platform

recommended

Usage

Per-user-per-month

  • ·Unlimited inferences
  • ·Custom retention windows
  • ·SLA + dedicated support
  • ·Workflow integrations

Enterprise

Contact

Employer · payer · clinical

  • ·BAA / DPA / SOC 2
  • ·Custom signing keys
  • ·On-prem inference option
  • ·Procurement support
--The 100M question

From 5.5M to 100M in 18–24 months. The architecture, by milestone.

Hardware can't add 95M users in 24 months. The only motion that delivers users in millions per deal is concentrated B2B — payers, employers, pharma. The window is set by Apple's HealthKit rollout. See what Continuum looks like at 10M, 50M, and 100M.