What changes with Ariadna
Complex investigations often mean opening registries, crossing names, rebuilding timelines and trying not to lose where each fact came from. Ariadna turns that scattered work into a structured investigation that preserves evidence and separates facts, hypotheses and missing information.
Multi-source investigation system that reconstructs structures, timelines and relationships, measures how atypical a configuration is and preserves auditable evidence without turning anomaly into accusation.
What it can address
- Corporate and registry investigation
- Timeline reconstruction
- Hypotheses and alternative explanations
- Auditable dossiers
Evidence-based investigation, not classification of people
Ariadna starts from an entity, person or company and reconstructs the corporate and temporal structure observable in public records. Its purpose is not to assign guilt or produce a probability of criminal conduct. The system presents facts, inferences, contradictions, missing data and legitimate explanations that could or could not be ruled out.
Three engines with separate responsibilities
CASANDRA DDE expands the investigation and discovers candidates, relationships and new sources. CERBERUS DRE verifies and refutes claims reproducibly. METIS ADE can, on request, evaluate whether the assembled evidence justifies human review. Ariadna provides domain logic, connectors, criteria and user experience while the engine cores remain separated.
Search-bias control
Architecture prevents a model from deciding by itself which evidence is allowed to exist. A rules-based baseline expansion runs before model prioritization; hypotheses must search for confirming evidence, evidence capable of destroying them and observations that discriminate between rival explanations. Uninvestigated eligible links remain recorded.
Contextual rarity, not adjectives
Patterns are compared against groups of similar organizations and expressed in understandable bands together with population size and coverage. An atypical configuration does not automatically become a conclusion: it triggers review of alternative explanations and evidence quality.
Reproducible timeline
The temporal sequence answers questions a static graph cannot: when a person appeared, which relationship existed before or after a change and how the structure evolved. Every relevant element should open its source, date and supporting fragment.
The auditable dossier is the deliverable
The panel is the working interface; the dossier is the defensible artifact. It includes versions, sources, claim classification, strong or weak identity, statistical comparison, alternative explanations, coverage, truncations, declared hypothesis, missing data and an explicit scope warning.
Varexis as the sensitive-data boundary
When external models are involved, personal data passes through Varexis before leaving the controlled environment. The model can preserve relationships through opaque tokens without necessarily receiving real names or identifiers, and each protection operation can leave verifiable evidence.
Declared limits
Ariadna does not prove tax evasion, does not determine criminal association and cannot exceed the limits of public sources. Captchas, credentials, payments or legal authority trigger human intervention. The system is designed to declare those limits rather than hide them behind a score.
Enter through the problem, not the technology.
This minisite separates experience, operations and engineering so each reader can go only as deep as needed.
Access and licensing
Free usage licences for selected non-commercial institutions and public bodies, including ARCA. For-profit use requires a paid commercial licence. Source-code access follows explicit terms.