Home Pillars Our AI Competition Why We Win Scoring Pricing Blog Our CEO Join Founders Club — $25 →
AI Engine

Our AI

The intelligence layer that turns raw biometric data into 30-day predictive health forecasts.

How It Works

From Data to Prediction

Hunuu Sense AI™ ingests data from 50+ wearables, normalizes it against your personal baseline, and surfaces actionable health intelligence — continuously.

📡
Data Ingestion

Real-time streams from 50+ wearables — Apple Watch, Oura, Whoop, Garmin and more — normalized into a unified data schema.

🧮
Contextual Baseline

A tennis player's heart rate is not a mall walker's. Our AI learns YOUR normal — not population averages.

🔮
Predictive Engine

Pattern recognition trained on 100M+ data points. Detects physiological anomalies 30 days before symptoms appear.

💬
Plain Language Output

No PhD required. Complex biomarker analysis delivered as clear, actionable guidance you can actually use.

🏥
EHR Integration

Bidirectional integration with Epic and Cerner. Your wearable data meets your clinical record.

🔒
Privacy by Design

HIPAA compliant. End-to-end encrypted. Your data is never sold. You own it — always.

// Developer Infrastructure

9 Live APIs.
One Intelligence Layer.

Hunuu isn't just a consumer platform — it's infrastructure. Our API layer powers the HIP App, the SIQ Score engine, and B2B white-label deployments for health systems. Every insight we surface is grounded in real-time clinical data.

🏥
NPI Registry
Every provider verified in real-time against the National Provider Identifier database.
🤖
Claude AI
Natural language intelligence layer — translates complex biomarker patterns into plain-English insights.
📡
Wearable APIs
Apple HealthKit, Google Fit, Oura, Garmin, WHOOP — all normalized into one signal.
💊
FDA Drug
Live FDA drug database for supplement and medication cross-referencing against your biomarkers.
⚖️
CourtListener
Malpractice and disciplinary records power the SIQ Score — no high-risk provider passes undetected.
📚
PubMed
Every health insight grounded in peer-reviewed research from the National Library of Medicine.

+ RxNorm · Stripe · SIQ Score API (Beta)

Full API Documentation →
// Predictive Intelligence Layer

Why We Use
TensorFlow

TensorFlow isn't a buzzword in our stack — it's the engine that makes 30-day prediction possible. Our LSTM (Long Short-Term Memory) neural networks are trained on multi-parameter biometric sequences: HRV, cortisol rhythm, glucose variability, SpO2, skin temperature, and sleep architecture — simultaneously.

LSTM is purpose-built for time-series with memory. It doesn't just look at today's HRV — it understands the pattern of your HRV over 30, 60, 90 days and detects when the trajectory is drifting toward a health event before any symptom appears. Retrospective validation: 94.2% accuracy across n=847 cases.

🧠
LSTM Neural Networks
Multi-layer recurrent architecture trained on 12 concurrent biomarker streams. Captures long-range temporal dependencies that simpler models miss entirely.
📐
Dynamic Personal Baseline
Models calibrate to your biology — not population averages. A 45-year-old athlete and a 45-year-old sedentary person have different normals. Hunuu knows the difference.
🎯
Implicit Health Demand Signals
IHDS detection fires when correlated anomalies across 3+ biomarker streams exceed threshold drift. The system identifies demand for healthcare intervention before the patient or doctor does.
94.2%
Prediction Accuracy
Retrospective validation · n=847
30
Day Horizon
12
Biomarker Streams
50+
Device Sources
0.1ε
Privacy Budget
vs. Standard PostgreSQL
TimescaleDB Query Time ~8ms
Standard PostgreSQL ~3,200ms
Querying 90-day HRV at 1Hz · Billion row dataset
400×
Faster queries
10B+
Data points handled
// Time-Series Infrastructure

Why TimescaleDB
Changes Everything

Health data is fundamentally time-series data. Your HRV at 7:14am on Tuesday matters differently than the same HRV reading at 11pm on Friday. Standard databases — PostgreSQL, MySQL, MongoDB — were never built for this. They degrade catastrophically at scale with time-ordered, high-frequency data.

TimescaleDB is PostgreSQL extended specifically for time-series at billion-row scale. Combined with pgvector for AI embedding storage, Hunuu can query your 90-day biometric history in ~8ms — fast enough to power real-time clinical dashboards, live AI inference, and instant SIQ Score lookups simultaneously.

Continuous aggregates — pre-computed rollups mean your 12-month trend loads instantly, not after a 3-second full-table scan.
Automated data tiering — recent data stays hot (fast SSD), older data compresses automatically. 90% storage cost reduction vs. flat PostgreSQL.
pgvector integration — AI embeddings live alongside time-series data in the same database. TensorFlow model outputs stored and queried in microseconds.
Enterprise-grade reliability — full ACID compliance, replication, and backup built in. No custom sharding hacks required at scale.
// Security Architecture

Network. Service. Application.
Three Layers. Zero Compromise.

Health data is the most sensitive data that exists. We built security into every layer of the stack — not as compliance checkbox, but as product architecture.

🌐
Network Layer

All traffic encrypted in transit via TLS 1.3. Caddy reverse proxy with automatic HTTPS provisioning. Zero plaintext connections permitted — enforced at infrastructure level, not application level.

TLS 1.3 end-to-end encryption
HSTS with 1-year max-age
Caddy auto-renewing SSL certificates
DDoS mitigation at edge layer
Zero-trust network architecture
⚙️
Service Layer

Every API call authenticated via JWT with short-lived tokens. Stripe webhook signatures verified on every payment event. Rate limiting, request signing, and audit logging at every service boundary.

JWT authentication — 15-min token expiry
Signed Stripe webhook verification
API rate limiting per key/IP
Full audit log — every data access recorded
HIPAA-compliant API endpoints
🔐
Application Layer

AES-256 encryption at rest for all PHI. Differential privacy (ε ≤ 0.1) prevents reverse-engineering individual records from aggregate AI models. Health Circles enforce patient-controlled data sharing at the field level — not just the record level.

AES-256 encryption at rest (all PHI)
Differential privacy ε ≤ 0.1
Health Circle 4-tier access control
Zero data sold to insurers, pharma, or gov
SOC 2 Type II — in progress
Compliance Status
Security is not a product feature. It is the product.
HIPAA Compliant EFF Certified Patent Pending SOC 2 In Progress AES-256
// Adaptive Interface

The Mood Engine

No other health platform does this. The HIP interface shifts its own visual state — color palette, typography, even animation rhythm — based on your live biometric signals. The app reads your body, then meets you where you are.

// Inputs to the Engine
HRVAutonomic state
CortisolStress curve
Sleep QualityOvernight recovery
ActivitySteps, workouts
HormonesTestosterone, estrogen
Time of DayCircadian phase
Day of WeekWeekly rhythm
Spiritual CalendarSunday, fasts, rites
// 12 Mood States

Your Biology, Your Interface

01 · OPTIMAL

HRV >70, Mood >80, Sleep >7h. Electric wasabi on teal. Peak performance.

02 · ENERGIZED

Steps >10k, elevated HR, post-caffeine. Amber kinetic.

03 · CALM

Low resting HR, high HRV, evening. Ocean stillwater.

04 · FOCUSED

Sustained HRV, flow state, work hours. Ice blue laser.

05 · RECOVERY

Post-workout, low glycogen, rest day. Healing amber.

06 · STRESSED

HRV <40, cortisol spike. Magma red alert. Breathwork nudge.

07 · SOCIAL

Evening 6–11pm, weekend. Warm purple + gold.

08 · CREATIVE

Alpha waves, studio hours. Prismatic rainbow shift.

09 · SPIRITUAL

Sunday, prayer >20min, Faith IQ >85. Sacred indigo + gold.

10 · TIRED

HRV <35, Sleep <5h, Monday AM. Muted navy. Reduces visual noise.

11 · ROMANTIC

Partner proximity, Sex IQ >85, weekend night. Deep rose + champagne.

12 · WARRIOR

Peak T, competition day, high cortisol+energy. Blood red + tactical black.

This isn’t theming. It’s adaptive cognition. A stressed user gets a stressed UI — and prompts calibrated to ground them. A warrior gets tactical. A spiritual user gets sacred. The software sees you.