Corpus Protocol
Decentralized BioData & Health Protocol for the AI Era
A consent-based way for people to own their biological and health data — and for AI to learn from it responsibly, without ever touching the raw file.
No spam. Just early access & launch updates.
AI's next breakthrough needs your biology. It shouldn't need your consent to be optional.
Modern AI models for drug discovery, biomarker prediction, and personalized medicine are only as good as the biological data they train on. That data — genomes, wearable signals, clinical histories — is some of the most sensitive information a person has, and today it mostly moves through centralized companies with little say from the people it belongs to.
Corpus Protocol is designed around a simple principle: your raw data never has to leave your hands for it to be useful. Instead of handing files over to a company, your data stays in an encrypted vault under your own key. Researchers submit questions, not data requests — and those questions are answered inside secure, isolated compute environments that return only an aggregate result, never the underlying record.
Consent stays granular and reversible: you decide which categories of research your data can support, and you can revoke that access at any time. Nothing is sold. Nothing is shared by default. The goal is a model where AI progress and personal data sovereignty aren't in conflict — and where the people supplying the data that makes discoveries possible are fairly recognized for it.
Three ideas doing the actual work behind the concept.
Decentralized Bio-Vaults
Your genomic, wearable, and health data is encrypted and stored in a personal vault secured by a key only you control — not the protocol, not a company, not a third party.
- Data is encrypted before it ever leaves your device
- No central database of raw records to breach or sell
- Granular, revocable consent per research category
Zero-Knowledge AI Training
Research queries run inside Trusted Execution Environments — sealed, hardware-verified compute where a model can learn from your data without your data ever being exposed, even to the operator running it.
- Only statistical results leave the enclave, never raw files
- Every result carries a hardware attestation proof, on-chain
- Query logic is auditable — trust doesn't require blind faith
Fair Value Redistribution
When your data contributes to a research result, value flows back to you directly — not to a middleman holding your file. Contribution is tracked transparently, and payouts settle automatically.
- Majority of every query's value goes to data contributors
- Payouts settle on-chain — no invoicing, no waiting
- No token speculation required to participate
Corpus Protocol is still early. The waitlist is where it starts.
Join now for early access when the vault opens, and follow along as the protocol is built in the open.