Quantum threat is real

Post-Quantum Authentication

RSA and ECDSA fall to a sufficiently powerful quantum computer. H33 replaces them with three independent PQ signature families, FHE-encrypted biometrics, and zero plaintext exposure — in a single API call.

1.36ms

Full Pipeline

42μs

Per Auth

PQ Families

0 bytes

Plaintext Exposed

Get API Key Read Docs

The problem

Classical Authentication Is Already Broken

Every RSA key, every ECDSA signature, every TLS handshake protecting your users today relies on one mathematical assumption: that factoring large integers or computing discrete logarithms is hard.

A sufficiently powerful quantum computer solves both in polynomial time. Shor's algorithm does not need to be fast — it just needs to exist. And it will.

Harvest-now, decrypt-later: adversaries are recording encrypted traffic today, waiting for quantum capability to decrypt it. Your authentication tokens, session keys, and signed credentials are already at risk.

NIST finalized post-quantum standards in August 2024. The migration deadline is not theoretical. It is policy.

The solution

Three Independent Mathematical Bets

H33 does not gamble on a single post-quantum scheme. Every authentication is signed under three independent PQ families:

ML-DSA-65 (Dilithium)

Module-lattice digital signatures. NIST FIPS 204. Hardness: MLWE lattices.

FALCON-512

NTRU-lattice signatures with compact keys. Hardness: NTRU lattices (independent of MLWE).

SLH-DSA-SHA2-128f (SPHINCS+)

Stateless hash-based signatures. NIST FIPS 205. Hardness: hash function preimage resistance.

An attacker must simultaneously break MLWE lattices, NTRU lattices, and stateless hash functions. Three independent mathematical bets, one authentication.

How it works

Three Steps. One API Call.

Biometric data is encrypted client-side via FHE. The server computes a match on ciphertext. The result is attested under H33-74 with three PQ signature families.

Step 1

Encrypt

Biometric template encrypted client-side using BFV FHE. 32 users batched per ciphertext via SIMD slots.

→

Step 2

Authenticate

FHE inner product computed on encrypted data. ZK-STARK lookup verifies result. Server never sees plaintext.

→

Step 3

Attest via H33-74

Result committed to 74-byte substrate: 32 bytes anchored on Bitcoin, 42 bytes in Cachee. Three PQ signatures.

        // Authenticate a user with post-quantum attestation

        POST /v1/auth/verify

        {

          "template": "<base64-fhe-ciphertext>",

          "enrolled_id": "user_29x8k",

          "attest": true

        }

        // Response

        {

          "match": true,

          "confidence": 0.9847,

          "h33_74": "<74-byte-substrate>",

          "pipeline_ms": 1.36,

          "signatures": ["ml-dsa-65", "falcon-512", "slh-dsa-sha2-128f"]

        }

Performance

Production Numbers, Not Projections

Benchmarked on Graviton4 metal-48xl (192 vCPUs). Full FHE + attestation + ZKP pipeline. Not synthetic microbenchmarks.

1,667,875

Sustained auth/sec (30s, full pipeline)

42 μs

Per-authentication latency

1,345 μs

32-user batch (FHE + attest + ZKP)

Stage	Component	Latency	% Pipeline
1. FHE Batch	BFV inner product (32 users/CT)	943 μs	70%
2. Batch Attest	SHA3 + Dilithium sign + verify	391 μs	29%
3. ZKP Lookup	CacheeEngine cached verification	0.358 μs	<1%

Post-Quantum Authentication

Classical Authentication Is Already Broken

Three Independent Mathematical Bets

FHE Overview

ZK Proofs

Biometrics

H33-74 Whitepaper

Pricing

Benchmarks