The one-paragraph version
Your phone captures and encrypts your identity evidence. A decentralised network of independent computers verifies it and records tamper-proof results — never the evidence itself. When a service needs an answer about you, cryptography delivers the answer without the data. No single company holds a database of everyone's documents.
The pieces, named
| Component | What it is | What it does for you |
|---|---|---|
| Identity Wallet app | Open-source mobile app (the reference capture app) | Guides document scan, selfie and liveness challenges; encrypts everything on your device before transmission |
| VEID | The identity layer of the VirtEngine blockchain | Manages your encrypted identity scopes, consent records, verification levels and proofs |
| VirtEngine network | An open-source decentralised network (CometBFT / Cosmos SDK), written in Go, Apache 2.0 licensed | Independent validators verify identity claims and keep tamper-proof records — no central database |
| Envelope encryption | X25519-XSalsa20-Poly1305, applied on your device | Your evidence is ciphertext everywhere except your own phone |
| Zero-knowledge proofs | Cryptographic proofs (the network's zk module) | Prove statements — “over 18”, “meets level Standard” — without revealing the underlying data |
| Device attestation | Google Play Integrity / Apple App Attest | Proves the app and device are genuine and untampered for high-trust verification |
Why a blockchain is involved
Three properties are hard to get any other way. No central honeypot: there is no single database of everyone's documents to breach. Tamper-proof records: once the network records a verification result, no administrator can quietly edit it. No gatekeeper: no single company can decide to revoke everyone's identity or sell access to it.
Personal data is never written to the chain in readable form — only encrypted references and verification results. See deleting your identity for how deletion works in this design.
The machine-learning verification
Document authenticity, face matching and liveness are assessed by machine-learning models. The method — combining document capture, biometric liveness verification and hardware attestation into a single verifiable pipeline — is what Australian patent AU2024203136B2 covers. Verification runs in controlled environments; what gets recorded is scores and results, not your images.
Want the engineering depth?
Full protocol documentation lives at docs.virtengine.com, and the complete source is on GitHub — including the x/veid module and the mobile capture app this site describes.