Aggregate. Analyze. Predict.
Lumus is a modular intelligence platform that unifies every available data source into a single operational interface — institutional databases, HUMINT, SIGINT, OSINT, GEOINT, FININT — eliminating the fragmentation between systems that currently forces analysts to manually search for information across separate platforms. One query, all sources interrogated in parallel, results aggregated in seconds.
The system creates a unique card for each entity (person, company, account, vehicle, device, any investigative object), eliminating duplicates and automatically reconstructing all connections: relationships, transactions, movements, communications, properties. Entity resolution recognizes the same entity even under different names, aliases or identifiers.
Beyond aggregation, Lumus applies automatic correlation algorithms and AI models to perform thematic screening (counter-terrorism, organized crime, money laundering, fraud, cybercrime), build relational graphs, detect temporal anomalies and generate quantified risk assessments. The predictive component identifies threats before they materialize, projecting future scenarios based on historical patterns and weak signals.
Technical Architecture
Lumus is a web application accessible via browser on authorized workstations. The interface is lightweight: all computational work takes place on dedicated central servers, hosted within government infrastructure, where the AI engine, relational graphs, correlations and continuous monitoring reside. The analyst installs nothing locally — they log in, search, receive results.
Existing databases are neither touched nor duplicated. Lumus connects to them through a network of secure APIs that query each source in read-only mode, translating different formats and protocols into a common language. Each agency keeps its own systems: Lumus virtualizes them into a unified view. Integrating a new source simply requires developing a dedicated API connector, without modifying anything in the existing infrastructure.
Eight Operational Modules
Live Broad-Spectrum Search: Simultaneous interrogation of all available sources with unified results on a single page. One search, every database, immediate results.
Continuous Monitoring: Real-time automatic alerts on any relevant change for monitored entities: transactions, movements, corporate changes, new properties.
GeoIntel Map / Live Heatmap: Advanced geographic visualization overlaying events, properties, utilities, territorial movements and anomalous activity concentrations on interactive maps.
Net Device Intel: Correlation between devices, SIMs, IPs, cell towers and network access points. Identifies who uses what and where, even when devices or SIMs are swapped.
Transport Track: Complete reconstruction of physical movements: vehicles, transits, checkpoints, cross-border passages. Identifies co-locations and meetings between different subjects.
UBO Universe: Automatic reconstruction of multi-level beneficial ownership chains. Traces the real control behind shell companies, nominee directors and corporate pyramids.
Social Contact Sieve: Contact network analysis between subjects. Filters out noise, identifies significant connections, central nodes, intermediaries and relational clusters.
Recognizer: AI pattern recognition on unstructured content: handwriting, texts, images, photos, behaviors. Transforms qualitative data into structured, correlatable intelligence.
Each module can be activated independently and shares data with the others. The architecture is scalable and can be implemented progressively without rewriting the system.
The technical dossier and MVP are available upon request.
mattia@msztcapital.com