Joint Technical Paper
PayKing Corporation / KoreaGemLab × H33.ai, Inc. · May 2026
Provable Continuity

Toward a Fully Attested Ownership Substrate

When trust fails, proof survives.

Five layers that make institutional memory survive failure. Across time, systems, vendors, and cryptographic eras. The ownership object is the proof. They are inseparable.

Read the Paper Verify a Proof Explore the Architecture
Seong-Ho Lee × Eric Beans
The Core Problem

Modern systems cannot prove what actually happened once trust breaks.

When a $12 million wire transfer is disputed eighteen months after execution, the institution must answer a simple question: what happened?

In most systems today, the answer depends on whether logs survived, whether vendors cooperated, whether databases remained intact, and whether the AI model that made the compliance decision still exists in its original form.

When any of these assumptions fail — and they fail precisely when the evidence matters most — the institution cannot prove what happened.

This is not a security failure.
It is a continuity failure.

Canonical Incident — The Wire That Cannot Be Proven

An AI-assisted compliance system approves a $12M wire transfer. Six months later, the transfer is disputed.

Without Continuity Infrastructure

  • SIEM logs incomplete (90-day retention expired)
  • AI model updated twice since the decision
  • Compliance vendor changed platforms
  • Screenshots disputed as fabricated
  • Database records modified during investigation
  • 3-6 week forensic reconstruction attempt
Cannot prove what the AI saw, why the decision occurred, or whether the record changed.

With Fully Attested Continuity

  • Original ownership state reconstructed
  • State transition attestation verified
  • Governance approval proof replayed
  • AI decision independently reproduced
  • Tampering detected cryptographically
  • Offline proof succeeds — no vendor, no platform
Every action reconstructed. Every policy replayed. Every record verified. Independent of all infrastructure.
The Consequence of Inaction

What failure looks like without provable continuity.

AI decisions cannot be replayed — the model, the policy, and the input are gone
Ownership breaks permanently after key loss — no recovery path exists
Vendors become evidentiary dependencies — if they disappear, so does proof
Governance approvals become disputable — logs are not cryptographic evidence
Audit reconstruction takes weeks and produces inconclusive results
Cryptographic transitions invalidate entire trust chains retroactively

Provable continuity eliminates evidentiary discontinuity.

What changes when truth becomes provable.

Logs expireReplay survives
AI model changesDecision preserved
Vendor dependencyIndependent verification
Screenshots disputedCryptographic reconstruction
Reconstruction: weeksVerification: minutes
Ownership: key-dependentOwnership: identity-permanent
TrustProof
The Architecture

Each layer answers a question that regulated systems cannot answer today.

Five layers. One evidentiary chain. Together they produce a complete ownership and governance proof that survives time, infrastructure changes, vendor disappearance, and cryptographic era transitions.

1

Legal Ownership Foundation

Who owned the asset?

Legal identity committed on-chain without exposing personal data. Ownership proof survives key loss, chain migration, and intermediary removal. The foundation upon which all other layers depend.

2

Attested State Continuity

Was the state transition legitimate?

Every ownership change produces a post-quantum attestation. State transitions are cryptographically bound to the prior state. The attestation chain is tamper-evident and independently replayable.

3

Governance Attestation

Who authorized it?

Multi-party governance proofs bound to every state transition. Quorum thresholds are policy-configurable and independently verifiable. Each approver's authority is provable at the governance event timestamp.

4

AI / Agent Governance

Why did the AI approve it?

The AI decision becomes evidence. The input, the model version, the active policy, and the output are cryptographically bound into a single attestation at the moment of execution. Compliance is evaluated on encrypted data — the system never sees plaintext.

Eighteen months later, when the regulator asks what model version executed, which policy was active, and whether the input data was unchanged — the answer is not a log entry. It is a cryptographic proof that any party can independently replay. Without the vendor. Without the platform. Without the AI system itself.

This reframes AI governance from monitoring and alignment to permanent, replayable, cryptographic evidence.

5

Post-Quantum Survivability

Will the proof hold in 30 years?

Three independent post-quantum signature families — three independent mathematical hardness assumptions. If one family weakens, two remain independently secure. Evidence survives cryptographic era transitions without reissuance.

No vendor required.No platform required.Verification independent of trust.

Any party — regulator, auditor, counterparty, court — can independently verify any state transition. Verification does not depend on vendor cooperation, platform availability, or institutional trust. It depends on mathematics.

Verification is offline — no API call, no platform access, no vendor cooperation
Same inputs always produce the same verification result for any party
Reference verifier is open source — standalone binary, no dependencies
Attestation chain is independently replayable at any past timestamp
Three PQ signature families — three independent hardness assumptions
Chain-agnostic — proofs migrate across blockchains without reissuance
Consequences

Where continuity changes the outcome.

Tokenized Securities

Legal ownership bound on-chain. Survives key loss, custody changes, and chain migration. CLARITY Act aligned.

Banking & Settlement

Wire approvals, compliance decisions, and settlement events become independently reconstructable evidence.

AI Governance

AI compliance decisions become cryptographic evidence — replayable, version-bound, and provable years later.

Cyber Insurance

Governance state at time of incident is deterministically reconstructable. Claims become provable, not argued.

Regulatory Audit

Full event continuity replaces sampling. Every action is attested, cached, and replayable — not a statistical subset.

Cross-Border Payments

Counterparty identity, governance approval, and compliance proof travel with the transaction. No intermediary dependency.

The industry has spent a decade building systems that record what happened.
This paper describes infrastructure that proves it.
Read the Paper

35-Page Technical Architecture

Five layers. Verification protocols. Canonical workflow. Post-quantum survivability analysis. CLARITY Act alignment.

Download PDF
Joint technical paper — PayKing Corporation / KoreaGemLab × H33.ai, Inc.
Patent pending: KR 10-2026-0068484
Authors
Seong-Ho Lee

Founder & CEO

PayKing Corporation / KoreaGemLab

Layer 1 — Legal Ownership Foundation

Patent: KR 10-2026-0068484

paykingos.com · seongho@paykingos.com

Eric Beans

CEO

H33.ai, Inc.

Layers 2-5 — Attestation, Governance, AI, PQ Survivability

7 Patents Pending · 300+ Claims

h33.ai · eb@h33.ai

Ownership continuity. Governance continuity. AI decision continuity. Cryptographic survivability. One incident. Five layers. Complete evidentiary reconstruction — independent of time, infrastructure, vendors, and cryptographic eras.

Provable continuity is not a feature. It is the minimum requirement for institutional-grade digital infrastructure.