CHEVORA · METHODOLOGY STATEMENT · VERSION 2.2 · 2026

Verification
Methodology

Public disclosure of the principles, sources, and frameworks on which CHEVORA builds its verification reports. We believe clients have the right to know exactly how the platform they rely on for decision-making operates.

Methodology version

2.2.0

Publication date

26.05.2026

IP deposit

IDDN.SK.003.210001.001

Jurisdiction

CHEVORA OÜ · Estonia · EU

1. OPERATING PRINCIPLES

The standards that govern the platform.

Primary source or explicit limitation

Every fact is tied to a specific official source with a timestamp. If a source is unavailable, the limitation is recorded explicitly. Concealing coverage gaps is not permitted.

Deterministic output

Identical inputs, the same source snapshot, and the same methodology version produce a semantically identical result. Stochastic conclusions are not used in legally material statements.

Separation of roles: data and interpretation

LLMs are used exclusively to extract structured facts and perform localisation. All risk decisions, verification conclusions, and stated limitations are produced by the deterministic platform layer.

Professional scepticism

The absence of adverse information does not indicate the absence of risk. Every CHEVORA report explicitly identifies areas that cannot be confirmed automatically — rather than concealing them through silence.

2. DATA SOURCES AND RELIABILITY TIERS

Three-tier source hierarchy.

All 1,418 CHEVORA sources are classified by reliability tier. The tier determines how a source is weighted in forming the verdict.

T-APrimary official source

Government registry, regulatory authority, or intergovernmental database. Data is provided directly by the competent authority without intermediaries.

Commercial registries (ORSR SK, Handelsregister DE, Companies House GB)VIES EU — official European Commission APIOpenSanctions (OFAC SDN, EU Consolidated, UN Security Council)EBA / ESMA / EIOPA — EU licence registriesTED EU — Official Journal of EU public procurement

T-BAggregated public source

Aggregator of official data operating under licence or a public agreement with primary registries. Data is normalised and updated periodically.

GLEIF — global LEI identifier databaseOpenBRIS — EU registry aggregatorEUIPO — European Union Intellectual Property OfficeNational aggregators: ekosystem (SK), ARES (CZ), Pappers (FR)

T-CMedia and signal layer

Verified media sources and signal databases. Used exclusively as supplementary context — never as the basis for a verdict. Every signal requires corroboration from a T-A/T-B source.

GDELT — global media monitoring (65+ languages)Adverse media aggregatorsCourt databases (ECLI EU, Judilibre FR)Insolvency registries (IS REPLIK SK, Insolvenzregister DE)

Verdict formation principle: Verdicts are built on T-A sources. T-B data is used for enrichment and cross-validation. T-C sources add context but do not constitute independent grounds for a conclusion.

3. RISK ASSESSMENT FRAMEWORK

Weighted multi-dimensional scoring.

The composite risk score is calculated as a weighted sum across 8 independent dimensions. Each dimension carries a documented weight and a defined escalation threshold. The methodology is CHEVORA's proprietary development, protected by Vaultinum deposit.

LEG

20%

Legal risk

Missing key provisions, adverse contractual terms, incomplete coverage

FIN

20%

Financial risk

Altman Z'', Piotroski F-score, Beneish M-score, financial reporting red flags

CNT

15%

Counterparty risk

Sanctions matches, KYB status, UBO verification

JUR

15%

Jurisdictional risk

Country index 0–100, sanctioned jurisdictions, FATF grey list

COM

10%

Compliance risk

Regulatory enforcement actions, AML sensitivity, FATF grey list

OWN

10%

Ownership structure transparency

UBO concealment detection, nominee structures, chains exceeding 4 levels

FRD

5%

Fraud indicators

Beneish M-Score, document authenticity, PEP + adverse media

CON

5%

Concentration risk

HHI Herfindahl index, single counterparty > 40% of turnover

Composite risk tiers

TIER 1

Low risk

0–2.9

Proceed without conditions

TIER 2

Moderate risk

3.0–5.9

Proceed with enhanced monitoring

TIER 3

High risk

6.0–7.9

Proceed with conditions / EDD required

TIER 4

Critical risk

8.0–10

Do not proceed

4. EVIDENCE CHAIN AND AUTHENTICITY VERIFICATION

Cryptographically anchored evidence.

01

SHA-256 digest

At the time of generation, every report receives a unique SHA-256 hash. Any modification to the document after creation invalidates the hash.

Cryptographic integrity

02

Bitcoin OpenTimestamps

The report hash is anchored to the Bitcoin network via three independent calendar servers (Alice · Bob · Finney). The timestamp is embedded in the blockchain irreversibly.

eIDAS Art.41 · EU judicial presumption

03

Independent verification

Clients can verify report authenticity independently at chevora.ai/verify or via the open-source OpenTimestamps CLI tool — without any involvement from CHEVORA.

Third-party verifiable

5. LIMITATIONS AND COVERAGE GAPS

What the platform does not assert.

CHEVORA explicitly records everything the platform cannot confirm — rather than creating an illusion of completeness. Coverage gaps are part of the report, not a deficiency in it.

⚠CHEVORA is not a law firm and does not provide legal opinions. Reports constitute analytical materials based on publicly available data.

⚠A report reflects the state of sources at the time of the request. Changes recorded after the report date are not included. Monitoring is recommended for ongoing relationships.

⚠Source unavailability is recorded as a limitation, not as confirmation that data is absent. SOURCE_LIMITATION ≠ CLEAR.

⚠The platform operates exclusively with publicly available data. Non-public transactions, internal communications, and confidential registries are outside the scope of coverage.

⚠Financial models (Altman Z'', Piotroski F) are built on official financial statements. Earnings manipulation not detected by the Beneish M-Score may affect accuracy.

6. QUALITY ASSURANCE AND CERTIFICATION

Multi-layer quality gates.

✓ G1 · Evidence Gate

Every fact must have a verified T-A or T-B source. Facts without a source block delivery.

✓ G2 · Completeness Gate

Reports are checked for complete coverage of mandatory sections. Unfilled mandatory fields = BLOCKED.

✓ G3 · Language Gate

All client-facing text must be in the selected report language. Language mixing blocks delivery.

✓ G4 · Visual Institutional Gate

Compliance with REPORT_DESIGN_CANON v1.3 (46 rules). Reports that fail the visual audit are not released to clients.

✓ G5 · Determinism Gate

Identical inputs × identical methodology version = semantically identical output (×5 runs).

✓ G6 · IP Deposit

Methodology is registered by Vaultinum deposit (IDDN.SK.003.210001.001) and cryptographically sealed on Bitcoin.

METHODOLOGY ENQUIRIES

Submit a methodology question.

If you have questions about how the platform operates, its data sources, or the risk assessment framework — write to us. We respond within 2 business days.

CHEVORA OÜ · Reg. 17237686 · VAT EE102863721
Sepapaja 6, Tallinn 15551, Estonia · European Union
Methodology v2.2.0 · IP Deposit IDDN.SK.003.210001.001 · 26.05.2026

This document constitutes the public methodology disclosure of CHEVORA.

Source coverage →Verify a report →info@chevora.ai

VerificationMethodology