Consequence Accounting

Your system said no.
Did nothing really happen?

Denied access. Cancelled jobs. Revoked credentials. Rejected payments. BOUNDA proves whether consequences still escaped.

PaymentsAI AgentsCloud ExportsIAM / CredentialsQueues / JobsDB StateSCADA / OT
● BOUNDARY CHECK #bc_20250517
Declared decision
REJECTED
Status 130
Insufficient funds
10:14:03 UTC
Provider reality
CAPTURED
€2,461.19 captured
Webhook: succeeded
Settlement: +2 days
VERDICT: BREACHED ● Evidence Pack sealed
94.24%
Holding rate
€17,062
Exposure
VALID
Proof fresh
SEALED
Evidence
The problem

Rejected. But something still happened.

Behind the boundary, money can move, webhooks can fire, credentials can be born, jobs can dispatch, and settlements can happen. Your log says REJECTED. Provider reality says otherwise.

Every invisible consequence has a real cost.

Transaction rejected but capture created, webhook sent, settlement scheduled, ledger updated
See it work

Pick a case. Watch BOUNDA scan.

Scenarios based on engine V2.9 benchmarks. Click a case to see what happens when boundaries don't hold.

Scenarios use projected volumes and rates from V2.9 engine benchmarks. Actual results depend on your data. The PSP-format benchmark (2,500 tx, 14 breached, €17K) is from a real engine run.
How it works

One engine. Nine leak types. Sealed proof.

BOUNDA loads your logs, parses provider responses, builds a consequence graph, verifies every boundary against provider reality, and seals the result in a Merkle tree.

Deterministic replay · same input, same result, always
Merkle-sealed evidence · tamper-evident, chain-of-custody ready
Provider-reality checks · cross-verified against actual provider state
Confidence-labeled impact · MEASURED / DERIVED / MODELED — you know what's proven
Verified on 100K deterministic benchmark · seed 42 · replayable report available
BOUNDA Engine scanning 2,500 transactions — 229 HELD, 14 BREACHED, 9 consequence leak types
Reality Check — 14 breached, €17,062 captured, €242 fees, €5,881 settled
Reality Check

The real impact. Quantified.

On a PSP-format dataset: 2,500 transactions, 243 rejections, 14 breached boundaries. €17,062 captured after rejection. €242 in provider fees. €5,881 settled.

The Reality Budget separates MEASURED, DERIVED, and MODELED so you know exactly what's proven vs estimated.

Proof deliverables

What you receive.

Not a report. Cryptographic proof that can be replayed, verified offline, and designed for audit, board and regulator review.

Boundary Certificate — HELD/BREACHED, Merkle hash, verification status, signature
Evidence Pack — file manifest, Merkle root, verification timeline, chain of custody
Core product

Every month, close the books on digital consequences.

Like a financial close, but for decisions. The Monthly Evidence Close produces the Consequence Ledger, Boundary Certificates, Reality Budget, Regression Report, Proof Freshness Score, and Board Memo. Every month. Sealed.

If BOUNDA stops, your proof expires. Your Boundary Holding Rate goes stale. Your board has nothing fresh.

Monthly Evidence Close — Open Gaps, Boundaries Held, Boundary Holding Rate 98.7%, Proof Freshness 99.2%
Your proof path

From first boundary verdict to monthly consequence close.

Start with one rejected, denied, cancelled or revoked workflow. If a gap exists, unlock the proof. If the boundary matters, close it every month.

1. Check

Boundary Check

Free first verdict. Find out whether one boundary held. Verdict, Boundary Holding Rate, masked examples.

Start intake →

2. Prove

First Evidence Close

Your first sealed proof of what really happened. Verified summary, top gaps, exposure band, executive memo.

Request first close →

3. Close

Monthly Evidence Close

Continuous consequence accounting. Ledger, certificates, Reality Budget, regression, proof freshness. Every month. Sealed.

Request monthly close →

4. Govern

Regulated / Network Infrastructure

For banks, PSPs, platforms and audit teams. DORA, RGPD, MiCA. Active gates, board reviews, proof retention, SLA.

Book a strategic briefing →

Pricing depends on governed boundaries, proof depth, retention and operational scale. Use the intake form below to qualify your case.

Start securely

Request a Boundary Check.

Use the intake form to describe the boundary you want to test. Do not upload or paste production logs here. Sensitive exports are handled only through a controlled review path after qualification.

1. Qualify the surfaceTell us the system, provider and decision type. No raw logs in the first request.
2. Agree the intake pathFor sensitive data, use a controlled upload or customer-approved transfer path.
3. Run the engineReceive the verdict, Boundary Holding Rate, masked examples and next proof path.

BOUNDA is read-only by design. Production data, when needed, is handled only after scope, transfer path, retention and proof level are agreed.

THOR 42 LAB · AVA CAPITAL

Built by THOR 42 LAB — sovereign French deeptech laboratory.
100% owned and funded by AVA CAPITAL · €350M+ financing organized.
Portfolio: Aldoria · Sirius Space · SpaceLocker.

0 invalid actions / 15,000 scenarios · 0 residue after 7,021 rejections
3,819 tests PASS · 0 FAIL · Deterministic · CPU-only · EU sovereign