Most health tools do one of these. VitaCrypt does all four — and the value is in how they compound.
A unified six-stream data fabric; every recommendation cited to live science; fully homomorphic encryption so our servers never see your plaintext; and personalization down to your genotype and local environment. Here is how each one works.
Unification doesn't happen on our servers. Your phone pulls each source, normalizes it into one schema, and only then encrypts the combined profile under your key. We only ever receive one ciphertext — never six plaintext feeds.
Wearables stream through Terra API; DNA, microbiome and lab files are uploaded; environment is pulled from your location.
Different formats — VCF, FHIR, CSV, JSON — are mapped into one queryable health schema in local storage.
The unified profile is encrypted with a key held in your iOS Keychain / Android Keystore. Plaintext never leaves the device.
| Source | What it adds | Availability |
|---|---|---|
| Genetics | Variants that reshape every other signal — folate cycle, lipids, caffeine, methylation | At MVP launch23andMe / Ancestry upload |
| Wearables | HRV, sleep stages, recovery, resting heart, activity load | At MVP launchTerra API · Apple Watch, Oura, Whoop |
| Environment | PM2.5, ozone, pollen, UV — by your location, in real time | At MVP launchPurpleAir / IQAir / OpenWeather |
| Lifestyle | Stress, mood, diet, sleep quality via NIH-validated PROMIS short-forms | At MVP launchin-app check-ins |
| Microbiome | Gut composition → inflammation, nutrient absorption, diet response | Upload at MVPpartner API on roadmap |
| Labs & blood | CRP, ApoB, HbA1c, lipid panels — the objective check on the wearable signal | Upload at MVPHL7/FHIR connectors 2027 |
See the six sources on the home page → Integration spec in the litepaper →
Most apps bake in whatever the guidelines said the year they shipped. VitaCrypt re-queries the literature continuously, so a recommendation can change the week a relevant paper, GWAS hit, or guideline update appears.
An agent continuously scans public literature and guidelines. It never touches your data — it works entirely on published papers.
Whether a finding applies to you is computed against your encrypted profile cryptographically — the model never sees your genotype or labs.
Each surfaced recommendation carries the source it rests on, graded by evidence strength.
We search PubMed (~40M indexed records) and the Cochrane Database of systematic reviews (~8,500 reviews), plus GWAS Catalog, ClinVar, and clinical guidance from USPSTF, WHO and NICE. "Cross-referenced" means we search these indexes for evidence relevant to your profile — not that 40M studies stand behind any one recommendation.
See the full corpus list → Matching protocol in the litepaper →
Fully Homomorphic Encryption (FHE) lets a server run analysis on data that stays locked the whole time. The result comes back encrypted; only your key opens it — like a sealed box a mathematician can still do arithmetic inside.
Encryption and decryption run client-side — the pattern Zama shipped in its iOS FHE SDK (2025). The heavy compute is centralized FHE on our servers, not on your phone. We're targeting the Zama Concrete ML / TFHE stack; multi-party blind computing (Nillion) is on the roadmap as defense-in-depth, not in the MVP.
Zama's published encrypted-DNA work (bounty #95, 2024) reached 96% accuracy on encrypted DNA ancestry classification — chromosome 22, 1000 Genomes — in about five minutes, with the query genome encrypted and the reference panel in clear. That's the honest scope today: ancestry-class inference, not whole-genome variant calling. It's the proof the primitive works at investigator-grade latency.
Architecture designed for HIPAA / GDPR alignment. Pre-MVP — no production health data yet.
Watch the live walkthrough → Threat model in the litepaper →
Generic advice optimizes for a median person who doesn't exist. VitaCrypt reads each signal against your genotype and local exposures — so the same air, the same HRV dip, the same lab value yields a different instruction for a different person.
Genotype + current exposures (air, season, location) + physiology (HRV, sleep, labs).
Gene × environment interactions are evaluated against the cited literature — under FHE.
A small set of concrete, time-bound steps — each with its evidence attached.
The folate-cycle variant blunts homocysteine clearance; ambient PM2.5 adds inflammatory load. Together they're linked to higher ischemic-heart and vascular risk.
ε4 carriers show a greater neuroinflammatory response to fine-particulate exposure — observed even in young adults — relevant to long-horizon cognitive risk.
David is a composite case from the litepaper. Values are illustrative; the gene–environment methodology and citations are real.
VitaCrypt opens to its first design partners in Q4 2026. Investors and prospective partners — reach out directly.