The layers
Records and case management — FIRs, case files, accused and complainant details. This layer is about completeness and process compliance, not analysis.
Tactical crime analysis — pattern and series identification, hotspot mapping, temporal analysis. The consumer is the operational commander.
Investigative analysis — support for individual cases: link analysis, communication analysis, financial trail reconstruction. The consumer is the investigating officer.
Intelligence management — structured handling of intelligence that is not evidence. This layer is systematically underbuilt.
Strategic analysis — longer-horizon crime trend analysis and performance analytics.
Data fusion and search — the connective layer: ingesting from multiple systems, resolving entities, and providing federated search.
Where units get stuck
The analytical layer is bolted onto a records system never designed for it. Entity resolution does not exist across systems. The intelligence layer is informal. Analytical capacity is thinner than tooling. Everything runs on exports.
What to build first
Fix ingestion for your three highest-volume source types. Establish entity resolution — without it, every subsequent capability degrades. Then add analysis.
Non-negotiable requirements
Deployment inside your own infrastructure. Audit logging at query level. Role-based access with purpose limitation. Exportable data in open formats.
On predictive tooling
Predictive policing is genuinely contested. A unit considering such tooling should be able to answer what the model is trained on, what it is forecasting, how it will be evaluated, and what oversight applies.
The realistic assessment
Most units do not need more software. They need their existing data to be ingestible, entities resolved, analysts trained, and access properly governed.
pi-scout provides the fusion and entity resolution layer for investigative and intelligence work, deployed on-premise with query-level auditing.
