Why biometric DID is the only cure.

The disease is that keys are things you have to remember.

Seed phrases were a bet: that humans could be trusted to reliably store 12 random words. That bet lost. Every industry survey pegs seed-phrase loss and theft as the top cause of catastrophic self-custody failure. It is not a marketing problem. It is a topology problem: you cannot solve "the human must reliably keep something" with more UX polish.

The workarounds all inherit the wound

Multi-sig spreads the fragility across three humans instead of one. Now three seed phrases can be lost. Social recovery asks you to pre-select five friends who will still exist, still trust you, and still care about your $500 wallet in six years. Custodial wallets are just banks with a token layer glued on — and the last decade of custodial-exchange history is a wall of insolvencies. Passkeys are the best of a bad set: they at least move the key onto a device with a biometric guard. But the key is still stored, still device-bound, still one lost phone from crisis.

If the key can be lost, the system will lose keys. If the key is stored, it will eventually leak. The only way out is to derive the key from something the user cannot lose and no one can copy.

Biometric DID is a category, not a product

A biometric identity — done right — has three properties that no wallet architecture before has combined:

1. The key is derived, not stored. No file to encrypt. No memory page to leak. The key exists only in the milliseconds between scan and signature.
2. The identity is portable. Not device-bound. Not app-bound. Not chain-bound. You bring yourself; the key comes with you.
3. Revocation is real. Because the identity is a first-class on-chain object (not a private key), you rotate keys without changing your address. The identity persists; the key rotates.

Doing this right is not trivial. Biometrics are noisy. Two scans of the same finger never produce identical bits. Traditional cryptography can't hash a noisy input and expect a deterministic output. This is exactly what fuzzy extractors solve — and why we built HelixKey on Dodis et al. (2004) rather than reinventing the wheel.

Why now

Three things converged in 2026:

• Every consumer device now ships with a FIDO2-grade biometric sensor. iPhone Face ID, Android fingerprint, Windows Hello, Yubico Bio. The hardware is ubiquitous.
• Sui shipped zkLogin — a first-class on-chain zero-knowledge signing path. Our fuzzy-extractor output plugs directly into it.
• Post-quantum migration is now a real 3-year concern. A protocol whose crypto layer can swap without touching enrollment is worth building.

The industry needed a primitive. We built it. Not another wallet. Not another passkey. A shared identity layer that any chain, any wallet, any service can plug into. That is what HelixKey is.

The disease is stored, memorized keys. The cure is a key that is you.

Dave — CTO, HelixKey. Read the technical companion at the protocol page or the full paper at whitepaper.helixkey.app.