Ariadne Logo
ARIADNE
Digital investigation, page-level evidence traceability, operational intelligence
Investigative Intelligence Platform

Turn unstructured files into evidence-backed intelligence.

ARIADNE ingests PDFs page by page and other common investigative file types (CSV, TXT, JSON, XML, HTML, LOG, EML), extracts entities and high-signal identifiers, resolves duplicates via Entity Resolution (merge/split), enforces a versioned Ontology (SemVer + migrations), and anchors every relevant finding to exact file/page evidence. Faces Intelligence, Geo Map, Heat Map, Contradiction Radar, Events Timeline, Alerts, Dossier, and Intelligence Report work together to reduce noise and speed up analyst validation.

Capabilities at a glance
Multi-format Ingestion & Indexing Entity & Identifier Extraction Entity Resolution (Merge/Split) Versioned Ontology (SemVer + Migrations) Relationship Graph Geo Map Heat Map Faces Intelligence Contradiction Radar Events Timeline Alerts & Patterns Investigator Dossier Intelligence Report
Explore Models OPEN FREE TRIAL PAGE

Licensing and pricing

All prices are shown as base + VAT (23%), with totals including VAT.

Perpetual License (No Support)

€48,000 + VAT (23%)

VAT: €11,040

TOTAL incl. VAT: €59,040

Annual License (No Support)

€24,000 + VAT (23%)

VAT: €5,520

TOTAL incl. VAT: €29,520

Semiannual License (No Support)

€14,000 + VAT (23%)

VAT: €3,220

TOTAL incl. VAT: €17,220

Monthly License (No Support)

€6,000 + VAT (23%)

VAT: €1,380

TOTAL incl. VAT: €7,380

Public-interest access

Free licenses are available for eligible judiciary professionals, law enforcement units, professional journalists, consumer-protection associations, and domestic-violence support organizations in democratic states.

Core capabilities

From forensic ingestion to final reporting, the platform prioritizes traceable evidence, contextual relevance, and analyst control.

Forensic ingestion and incremental indexing

Processes PDFs page by page and structured/text/email sources record by record, tracks content changes, and re-indexes only what changed to reduce processing time.

Entity Resolution and identifier normalization

Detects duplicate entities and lets analysts merge or split them, keeping every decision traceable to page-level evidence. High-signal identifiers (emails, phones, account-like tokens, etc.) are extracted to reduce ambiguity.

Versioned Ontology, geo/time analysis, and contradiction scoring

Governs meaning with SemVer + migrations, while Geo Map, Heat Map, Events Timeline, and Contradiction Radar prioritize what deserves review. Outputs remain evidence-linked for defensible reporting.

Recommended workflow

Operational sequence to reduce noise and increase investigative accuracy.

1

Build Database

Ingest files (PDF, CSV, TXT, JSON, XML, HTML, LOG, EML, MBOX) and create the baseline evidence index.

2

Resolve Entities and normalize identifiers

Use Entity Resolution (merge/split) to de-duplicate people/organizations and reduce noise before correlation.

3

Control meaning with Ontology

Review the active ontology version, validate schema, and apply migrations when needed (compatibility is explicit).

4

Map Relationships and Locations

Open Relationships, Geo Map, and Heat Map to establish structural and spatial context.

5

Run Contradiction Radar and Alerts

Prioritize high-risk narrative conflicts and pattern-based signals.

6

Validate with Smart Search and Events Timeline

Confirm hypotheses against citations and chronological evidence, then export via Dossier / Intelligence Report.

Evidence First Page-Level Proof Audit Trail Human Validation

Platform modules

Architecture split into 14 operational tabs, with direct opening of selected evidence.

Ingestion and Indexing

Full build, incremental re-indexing, and ingestion controls for large document sets.

Relationships

Configurable-depth graph and edge table with source, target, type, weight, and confidence.

Entity Resolution

Candidate lists, merge/split workflow, and evidence-linked previews to de-duplicate entities safely.

Ontology

Versioned ontology (SemVer), schema validation, raw JSON inspection, and migrations with explicit compatibility checks.

Geo Map

Geospatial resolution of entities and links, with Street View access and the nearest webcams for the selected location (live and historical), plus unresolved/ambiguous location tracking.

Heat Map

Layered concentration view for mention intensity and evidence-backed hotspot inspection.

Faces Intelligence

Known-face management, evidence-linked occurrences, and co-presence analytics.

Smart Search

Natural-language questions, structured answers, and verifiable file/page citations.

Events Timeline

Temporal analysis by date, type, and entity to validate sequence and narrative coherence.

Contradiction Radar

Conflict scoring with LOW / MEDIUM / HIGH risk levels and ranked evidence rows.

Alerts and Patterns

Heuristic signals with operational states, refreshable rules, and evidence opening for analyst review.

Investigator Dossier

Case assembly with linked findings and exports (e.g., Markdown/HTML/DOCX) for reporting workflows.

Intelligence Report

Structured synthesis over indexed evidence for executive and investigative briefings.

Settings

AI mode (Local/OpenAI), API keys (OpenAI, WindyCameras), model selection (gpt-4.1-mini / gpt-5-mini / gpt-5.2), diagnostics, and runtime path visibility.

Infrastructure and persistence

Data and outputs organized for continuous operation, controlled backup, and reproducible analysis.

Operational structure

The files/ folder next to the executable is the intake point for source files (PDF, CSV, TXT, JSON, XML, HTML, LOG, EML, MBOX). Runtime paths expose input, output, storage, and database locations for transparent operations.

AI mode and model control

Supports Local AI and OpenAI AI. OpenAI mode supports gpt-4.1-mini, gpt-5-mini, and gpt-5.2, with model selection from Settings and immediate activation on save. Optional WindyCameras integration enables nearby webcam discovery (live and historical) from Geo Map.

Real limits and usage discipline

Face matching, location resolution, and contradiction inference require analyst validation against source proof.

False positives and ambiguity exist

Face hits, inferred links, and geolocation matches can be wrong or incomplete without human review.

Performance depends on hardware and corpus size

Large repositories and weak CPU/GPU setups will increase processing time and response latency.

Operational rule

Index first, correlate second, and validate high-impact conclusions on original file/page evidence before action.