MICHAEL - 2026 Pricing, Features, Reviews & Alternatives

MICHAEL (Machine Intelligence Complementing Human Analysts on Every Lead) is an on-premises crime and intelligence analysis workstation for law enforcement agencies, fusion centers, real-time crime centers, fire and EMS dispatch, corrections, defense, academia, and corporate security teams. It installs directly on a Windows 10 or Windows 11 desktop, requires no server infrastructure, and runs fully offline on restricted or air-gapped networks with no external dependencies at runtime. All analysis executes locally on the analyst's own machine, and no agency data is transmitted to or stored on any vendor server.

MICHAEL consolidates 11 analysis views over a single loaded dataset, eliminating the need to export records into separate tools. Spatial analysis renders on a 3D globe with eight configurable layers including time extrusion, kernel density, 3D choropleth, graduated symbols, and a time scrubber. Temporal analysis covers six intervals and nine span methods. Twenty-three statistical modes spanning descriptive statistics, relationship tests, and financial analysis are computed entirely on the local machine with no external library or server call. Social network and link analysis includes six centrality metrics (degree, reach, reach efficiency, betweenness, closeness, and eigenvector) plus community detection by label propagation. Additional modules cover patrol workload analysis based on actual demand, unsupervised crime series detection, geographic profiling with buffer zone modeling and isoprobability contour lines, lexicon-based narrative search with proximity and regex syntax, a Gas Station Theory tool for video canvassing, person and vehicle search by physical description and partial identifiers, a live feed view supporting 14 source types, data enhancement with batch geocoding and nine computed fields, and export to i2 Analyst Notebook. Read-only connections to SQL Server, MySQL, PostgreSQL, and APIs are supported.

An integrated AI analyst answers questions about the loaded dataset specifically, writes and executes calculations against local records, builds charts and data grid filters from natural-language input, and proposes lexicon terms counted against loaded records before the analyst accepts them. The AI analyst connects via the analyst's own API key and routes directly to the provider with no intermediary server. MICHAEL discloses exactly what it sends before any transmission occurs, produces no threat scores or risk tiers, and traces every analytical method to published literature for explainable, briefing-ready output. Licensing is per machine with no user accounts, and all analytic modules are included at one price with no tiers or paid upgrades.