Confirm your day in seconds. Sync WHOOP, upload your bloodwork and your genetics. Every night a correlation engine tests all of it against itself — with the statistical controls to tell a real pattern from a coincidence.
Plenty of tools hold one of these. The value is in holding all four in the same rows, where they can be tested against each other.
A spreadsheet-style ledger. One row per day, one column per thing you track — habits, training, nutrition, weight — plus custom columns for whatever else matters to you.
WHOOP recovery, HRV, sleep stages, strain and workouts land in the same rows automatically. Calibration days and nights with sensor gaps are excluded from analysis rather than quietly averaged in.
Upload a lab PDF and its markers become a structured record with reference ranges and trends. The text is extracted in your browser, so the file — and the name and date of birth in its header — never leaves your machine.
A one-time methylation profile — MTHFR, COMT, MTR and the rest — read as a set of hypotheses about how you respond, never as a verdict. Your own bloodwork then confirms them or does not.
Every night, across every module at once. Known physiology gets tested against your data instead of assumed — and the engine goes looking for relationships nobody told it about.
A real result from a real 263-day record. Nothing seeded it — it surfaced in the nightly discovery pass. The same behaviour moves resting heart rate in the opposite direction (r = −0.25), which is exactly what you would expect if the effect is genuine.
Four guards stand between a coincidence and something the coach will tell you.
Controlled for time trend and sleep, so a relationship that is really just "both improved over six months" cannot present itself as cause.
Consecutive days are not independent samples. Significance is computed against an effective sample size, not a raw day count.
Test enough metrics against each other and chance alone manufactures hits. False-discovery-rate correction takes them back out.
Robust, probable, hypothesis. Anything you ask for on the spot is labelled a hypothesis no matter how strong the coefficient looks.
When textbook physiology does not hold for you, it is labelled contradicted rather than quietly dropped — which is usually the more useful finding. How the correlation engine works →
Grok 4.6 coaching grounded in your ledger, your WHOOP data, your bloodwork and every conversation you have had — so you always know where you stand and what to focus on next.
Confirm your day with natural conversation. Complete onboarding with your voice and talk to your coach hands-free, anytime.
Every confirmed day makes the connection more personal. Five intelligence layers — decay scoring, effectiveness tracking, confidence calibration, and memory consolidation — compound quality as your ledger grows.
Most AI apps quietly route you to smaller, cheaper models. We made the opposite choice on every single decision: quality first, everywhere.
Every answer — even a quick "what did I eat yesterday" — runs on Grok 4.6. There is no cheaper model in the stack. Member is more of your own history, thought about harder — not a prettier wrapper around a lesser model.
Your history is stored at the highest resolution the state of the art allows — full-dimension semantic memory, never truncated to save on storage. Every question re-ranks your entire relevant history with instruction-guided relevance before a single word is written.
Where other apps trim context to cut costs, we send more: your full conversation thread, your analytics, your learned patterns. Background analyses — mission coaching, journey narratives — run at maximum depth with zero time pressure.
Weight, workouts, nutrition, habits — confirm each day in a simple visual ledger with instant cloud sync.
Grok 4.6 coaching with streaming responses and goal projections.
Natural realtime voice conversations to confirm your day hands-free, anytime.
Temporal awareness and momentum detection that adapts coaching to your state.
Trigger analysis, personal records, and correlations across all your data.
PII encryption, prompt injection defense, rate limiting, and circuit breakers.
5-layer learning system with decay, effectiveness tracking, and automatic memory consolidation.
Structured mission setting with outcome, core goal, deadline, and 3 commitment pillars — your long-term focus, made explicit.
The nLight.fit MCP server, nLight.fit OpenAPI specification and nLight.fit developer resources are public. A read-only server with eighteen tools, running the same handlers the in-app coach uses. Start at the credential-free sandbox — no account required.
Try it now at POST https://nlight.fit/api/sandbox —
the same protocol, no credential. Production is
https://nlight.fit/api/mcp with a bearer token against
your own history — streaks, recovery context, bloodwork, the
correlation graph as raw edges.
{
"mcpServers": {
"nlight": {
"url": "https://nlight.fit/api/mcp",
"headers": {
"Authorization": "Bearer rt_ro_..."
}
}
}
}"What correlates with my HRV? Use the correlation edges, not the narrative."
A frontier model on every single request — continuously evaluated, never quietly downgraded
None of this works without your bloodwork, your genome and an honest account of your day. Here is exactly what we do with them — and what we don't.
Free accounts are open — no card, no waitlist. $50/year, tax included, funds frontier-model intelligence on your own record.
Coaching over your last 60 days at high reasoning. Tracking, WHOOP and charts are never metered.
Create Free AccountSign in first. Checkout is attached to your account so the membership cannot land on someone else.
Become a MemberCommitting to anything without tracking it is impossible.