AI-Powered Analysis

Your personal AI health analyst. Trained to read everything.

Upload your bloodwork, genetics, and wearable data. Refract's AI cross-references all of it — then builds you a precise, personalized plan for your supplements, diet, and training. Not generic advice. Your biology. Your plan.

AI-Generated Personal Plan Confidence: High
Supplements — Immediate Change
Replace folic acid with methylfolate 5-MTHF — 400mcg daily, morning
Your current B-complex contains folic acid. With your MTHFR C677T compound het genotype, folic acid blocks the very receptor it needs to work. Switch to 5-MTHF immediately. Also switch B12 to methylcobalamin sublingual.
Diet — Timing Adjustment
Move last meal to 7:00pm or earlier
Your MTNR1B G allele means eating late suppresses nighttime insulin secretion. Your CGM shows overnight glucose at 97 mg/dL — too high. Your Whoop deep sleep is averaging 44 min on these nights. Earlier eating will move both.
Training — Recovery Protocol
Stop cold plunge within 4h of strength sessions
You're doing ice baths immediately post-lifting. This is suppressing the mTOR signaling required for hypertrophy. Move cold to off days, or minimum 4–6h post-session. Your strength gains will respond.
Supplements — Add
Add magnesium glycinate 400mg — evening
Your serum magnesium appears normal but you're taking no magnesium supplement and your magnesium-dependent markers (COMT function, vitamin D activation, sleep quality) all suggest depletion. Glycinate form for sleep benefit.
Personalized Supplement Plan Diet Timing Optimization Training Load Analysis Genetic Variant Interpretation IV Therapy Confounder Detection Apple Health · Whoop · Oura · Garmin CGM Glucose Overlay Hormone Optimization Six-Month Roadmap Lab Confounder Detection Methylation Analysis Recovery Therapy Integration Personalized Supplement Plan Diet Timing Optimization Training Load Analysis Genetic Variant Interpretation IV Therapy Confounder Detection Apple Health · Whoop · Oura · Garmin CGM Glucose Overlay Hormone Optimization Six-Month Roadmap Lab Confounder Detection Methylation Analysis Recovery Therapy Integration
What Refract Delivers

Not a report. A plan.

The AI reads your data across every layer, then builds specific, ranked, actionable output for four things: what to take, what to eat, how to train, and what to track next.

01
Supplements

Your Exact Supplement Protocol

Not "consider magnesium." Your specific supplements, exact doses, optimal forms, optimal timing, and which of your current supplements are interfering with each other — or are the wrong form for your genetics.

Switch B12 from cyanocobalamin to methylcobalamin 1000mcg sublingual — FUT2 non-secretor absorbs oral poorly
Move vitamin D3 from morning fasted to largest meal — 40% absorption loss when taken without fat
Add copper 2mg — zinc at 35mg/day is actively depleting your copper stores
Stop folic acid — MTHFR C677T means it's blocking methylfolate at the receptor
02
Nutrition

Personalized Dietary Adjustments

Specific dietary changes based on your labs, genetics, and wearable patterns — not generic macros. Your APOE genotype changes how saturated fat affects your LDL. Your MTNR1B variant changes what time you should stop eating.

APOE e4/e3: cap saturated fat at 10% of calories — your LDL is disproportionately sensitive
Last meal before 7pm — MTNR1B + CGM data shows overnight glucose dysfunction from late eating
Increase dietary choline target to 700mg/day — PEMT variant + MTHFR means you can't synthesize enough
Add 3 Brazil nuts daily — selenium deficiency is impairing your T4→T3 thyroid conversion
03
Training

Training & Recovery Optimization

Your training load cross-referenced with your labs, HRV trends, and genetics. Are you recovering? Is your zone 2 distribution right? Are your recovery modalities helping or actively working against your goals?

Acute:chronic workload ratio at 1.6 — high injury risk window. Reduce volume 20% for 10 days
Ice bath within 1h of lifting is suppressing hypertrophy signaling — move to off days
72% of training time in Zone 3 — polarize: more Zone 2, more Zone 4/5, less middle
Ferritin at 28 is limiting your VO2 max ceiling — address iron before increasing training volume
04
Roadmap

Six-Month Action Roadmap

Sequenced interventions with specific retesting milestones. What to change first, what to add at month two, what to retest at month three, and what success looks like in the numbers.

Month 1: supplement form fixes + meal timing shift — no cost, high impact
Month 2: retest ferritin, hs-CRP, homocysteine — confirm response before next layer
Month 3: add targeted interventions based on retest — adjust training load if ferritin responds
Month 6: full panel retest — compare 12+ markers against baseline. Track what moved.
Powered by AI

Why AI makes this possible.

01
The cross-referencing a functional doctor does in their head — at scale
A skilled functional medicine physician mentally tracks 30–40 variables simultaneously. Refract's AI tracks hundreds — every marker against every other marker, your genetics, your history, your active modalities — in minutes. The question architecture that drives the analysis took months to build. It knows which variants change which interpretations.
02
Catches what humans systematically miss
Your doctor doesn't know you did a Myers cocktail 48 hours before your blood draw. They don't know your ice baths are blunting your hypertrophy. They don't know folic acid is wrong for your genotype. Refract's intake system captures all of it, flags confounders, and adjusts interpretation before a single finding is generated.
03
Gets smarter every time you upload
Quarterly uploads build a longitudinal picture. The AI tracks trends, not snapshots. Rising LH over three draws with stable testosterone means something different than a single low reading. Interventions get tested against their results. The plan evolves as your data does.
04
Functional medicine thinking. Not a chatbot.
Refract isn't "ask your labs a question." It's a structured decision engine built on the same logic framework that functional medicine physicians use — root cause analysis, systems thinking, genetic overlay, lifestyle cross-referencing. The AI operationalizes that framework across your specific data.
The Process

From upload to action plan in 48 hours.

01

Complete the intake

Medications, supplements with timing, recovery therapies, training load. This is what makes your labs interpretable. Takes 15–20 minutes.

02

Upload your data

Lab PDFs from any lab, 23andMe raw file, Apple Health export, Whoop/Oura/Garmin data, CGM export, Withings or smart scale CSV, Cronometer or MyFitnessPal nutrition export. Any combination — more layers means higher confidence output.

03

AI cross-references everything

Every marker is checked against every other relevant marker, your genetics, your context, your therapy history, and your wearable trends. Confounders are flagged automatically.

04

Receive your plan

Supplement protocol, dietary adjustments, training optimizations, and a six-month roadmap — all specific to your biology, not population averages.

05

Track what changes

Quarterly uploads track your markers over time. See which interventions are working in the numbers. The plan updates as your data does.

Data Sources

Everything feeds the analysis.

The more data layers you provide, the higher the confidence score and the more specific the plan. Start with labs — add layers over time.

Layer 01

Blood Work

Any PDF lab report from any lab. Standard panels or advanced markers — all parsed and interpreted against optimal ranges, not just reference ranges.

QuestLabCorp Any LabLongitudinal
Layer 02

Genetics

23andMe or AncestryDNA raw data file. 192+ clinically relevant variants analyzed — MTHFR, APOE, DIO2, COMT, FKBP5, ACTN3, and many more.

23andMeAncestryDNA Raw Data
Layer 03

Wearables

HRV trends, sleep staging, training load, heart rate zones, recovery scores. All major platforms supported with device-specific interpretation rules.

Apple HealthWhoop OuraGarmin
Layer 04

CGM Data

Continuous glucose monitor exports. Overnight patterns, postprandial spikes, reactive hypoglycemia — correlated with your genetics, sleep, and meal timing.

DexcomLibre LevelsNutrisense
Layer 05

Body Composition

Smart scale exports tracking weight, body fat %, muscle mass, visceral fat, and bone density over time. Longitudinal trends cross-referenced with labs, training load, and nutrition — because a number on a scale means nothing without context.

WithingsGarmin Index Fitbit AriaInBodyDEXA
Layer 06

Nutrition Data

Detailed dietary logs cross-referenced against your labs and genetics. Not generic macros — your actual micronutrient intake mapped against your deficiencies. Why is your ferritin low when you eat red meat? Why is your B12 borderline when you supplement? The answer is usually in the food log.

CronometerMyFitnessPal Lose It
Sample Output

This is what the analysis actually says.

Specific. Layered. Direct. No hedging. No generic advice. This is what your plan looks like.

refract_analysis_2026_q2.pdf — CONFIDENTIAL
Date
April 15, 2026
Data Layers
Labs · Genome · Whoop · CGM
Confidence
HIGH — 91/120
Tiers Analyzed
11 / 12
Priority Findings — AI Cross-Tier Analysis
Lp(a) + elevated homocysteine: compounding cardiovascular risk hidden in plain sight
High Priority

Your Lp(a) is 94 nmol/L — above the threshold where cardiovascular risk is independently elevated regardless of other lipid values. Your standard cholesterol panel looks unremarkable. This is exactly why Lp(a) is commonly missed. Simultaneously, your homocysteine at 13.2 μmol/L is elevated, and these two markers compound each other's risk. Your MTHFR compound heterozygous status is the driver — your methylation cycle is impaired and it's showing in the labs. Homocysteine is modifiable with the right B vitamin forms. The Lp(a) is largely genetic and requires a direct cardiology conversation.

Your supplement stack is actively working against your MTHFR genotype
Supplement Intervention

You are taking a B-complex containing 400mcg folic acid. With your MTHFR C677T compound heterozygous genotype, folic acid does not convert properly and actively blocks the methylfolate receptor — making your methylation problem worse, not better. Switch to 5-MTHF (methylfolate) 400mcg immediately. Your B-complex also contains cyanocobalamin — switch to methylcobalamin 1000mcg sublingual. These are not subtle upgrades. They are the correct forms for your specific genetics.

Thyroid conversion problem — normal TSH is masking it
Investigate

Your TSH is 1.8 mIU/L — your doctor flagged nothing. But your Free T3 is 2.6 pg/mL and your DIO2 Thr92Ala genotype is TT homozygous — meaning T4 to T3 conversion is impaired in brain and peripheral tissue. Your thyroid is producing hormone. The conversion is failing. Standard thyroid testing is designed to miss exactly this. Selenium (you are getting none in your current stack) is a required cofactor for this conversion. Your ferritin at 31 is also limiting thyroid peroxidase activity. Both are addressable before any medication conversation.

Ice bath timing is suppressing your muscle gains
Training Adjustment

Your intake shows cold water immersion within 30–60 minutes post-lifting, 4 days per week. Immediate post-resistance cold immersion blunts the mTOR and satellite cell signaling required for hypertrophy — this is well-established. The anti-inflammatory response you're trying to accelerate is the same response that drives muscle adaptation. Move cold immersion to off days, or minimum 4–6 hours after strength sessions. This change costs nothing and directly addresses a likely contributor to stalled strength progress.

MTNR1B + late eating + CGM: three signals pointing at the same problem
Diet Timing

Your MTNR1B G allele impairs nighttime insulin secretion. Your CGM shows overnight glucose consistently at 95–102 mg/dL. Your last meal averages 9:40pm. These three data points form a complete picture: late eating is causing a glucose spike during high melatonin, suppressing insulin secretion, and the elevated overnight glucose is showing up as reduced deep sleep (average 44 min on high-glucose nights vs. 71 min on low-glucose nights in your Whoop data). Moving your last meal to 7:00pm is a single change that addresses all three simultaneously.

CONFOUNDER DETECTED: Myers cocktail IV received 48 hours before blood draw. Your B12 result (1,847 pg/mL) and magnesium (2.4 mg/dL) reflect the infusion — not tissue status. These markers are excluded from dietary deficiency analysis. Flagged for redraw at 7-day clearance. All other markers unaffected.
Pricing

One-time reports or ongoing intelligence.

Start with a single analysis. Add data layers as you collect them. Upgrade to monitoring when you're ready to track change over time.

Follow-Up Analysis
$297 one-time

Quarterly re-analysis. See what's changed, what your interventions did to the numbers, and what to do next.

  • Updated lab analysis with trend comparison
  • Intervention response measurement
  • Updated supplement + diet plan
  • Recalibrated action plan for next quarter
  • Wearable and body comp trend review
Join Waitlist
Labs-Only Analysis
$197 one-time

No genetics or wearables yet. Start with your bloodwork — the most direct window into what's actually happening.

  • Full blood work analysis — all panels interpreted
  • Supplement timing audit + conflict detection
  • IV therapy confounder check
  • Personalized supplement protocol
  • Priority findings report
  • Missing data recommendations
Join Waitlist
192+
Genetic Variants
Analyzed
54
Health Systems
Cross-Referenced
64+
Lab Confounders
Auto-Detected
25+
Treatment Timing
Rules Enforced
48h
Report Delivery

Early Access

Your biology.
Your plan.

Tell me what data you have. I'll be in touch within 24 hours to walk you through what your report would look like and get you started.

What data do you have?

No spam. No sharing. Limited spots. $397 one-time, locked in for early access members.