Turn scattered signals into decision advantage.
Magen is S32’s AI-native, ontology-first intelligence platform. It fuses all-source data into a single, sourced picture — Gotham-grade depth, every claim cited, every action audited, on infrastructure you control.
The intelligence platform of the 2020s.
Legacy platforms made a trade you no longer have to: lock your data inside a proprietary ontology, wait months for vendor engineers to stand it up, and trust AI judgments you can’t inspect. Magen rejects that bargain.
Built on open, standard infrastructure, Magen gives analysts the depth of a mission-grade intelligence system — entity resolution, link analysis, geospatial and temporal reasoning, agentic AI — while keeping every fact traceable to its origin and every decision defensible to a commander, a court, or a coalition partner.
One platform, end to end
Intelligence Workspace
A live operational picture: a global map, two-speed search (keyword and cited AI answers), alerts, watchlists, and confidence-banded review queues.
Dossiers
Sourced subject files on people, organizations, vessels, aircraft, and facilities — every claim cited, with Analysis of Competing Hypotheses built in.
Conflict & War Monitoring
A war → conflict → battle → event hierarchy with per-zone escalation scoring, order-of-battle, course-of-action analysis, and temporal playback.
Investigations & Casework
A full case lifecycle — sources, evidence, claims, hypotheses, and tasks — on a spatial-temporal investigation canvas, fully audited.
AI & Autonomy
Agentic, citation-first retrieval across your data and the live web, plus durable background agents that propose findings, never decide them.
Geospatial & Imagery
PostGIS-backed mapping, areas of interest, STAC imagery ingest, detection review, range rings, and APP-6 event symbology.
Connectors & Ingestion
30+ open data feeds out of the box, geo-temporal and actor-overlap correlation, and an MCP server for structured machine ingest.
Records & Provenance
An ontology spine where every entity, claim, and value carries its provenance — raw, derived, enriched, or corroborated.
Reporting & Dissemination
Compile reports from sourced claims, evidence, and tasks, with every judgment referenced, then export to PDF, Word, Markdown, or JSON.
Why analysts trust what Magen tells them
Every fact carries its origin.
Magen is built on an ontology spine, not a pile of documents. Each value is tagged with how it got there, so analysts can see corroboration and contradiction at a glance — and stand behind every line of a report.
Provenance waterfall
- Raw As ingested from the source, untouched
- Derived Parsed, normalized, and structured
- Enriched Correlated and linked across sources
- Corroborated Confirmed by independent evidence
AI that cites its sources.
Magen’s agentic retrieval reasons over your data and the live web, then answers with numbered citations you can open. AI proposes; analysts decide. Nothing consequential happens without a human in the loop.
The accountability pattern
Background agents, on a leash.
Durable agents advance lines of inquiry and monitoring on their own — within hard spend rails, behind a kill switch, and with a full, replayable trace of every decision and tool call they made.
Guardrails on every agent
Magen vs. the black box
| Magen | Legacy platforms (e.g. Gotham) | |
|---|---|---|
| Architecture | Open stack — PostgreSQL, PostGIS, Neo4j, Qdrant, OpenSearch | Proprietary, closed ontology and pipelines |
| Your data | Stays in standard databases you can query, export, and own | ETL’d into a vendor black box |
| AI transparency | Every answer cited; consequential actions pass deterministic policy | Opaque model judgments, limited explainability |
| Audit | Append-only, policy-gated audit log on every mutation | Limited, platform-defined logging |
| Deployment | SaaS, on-prem Kubernetes, or air-gapped Docker — in days | Vendor-led onboarding measured in months |
| Lock-in | None — open formats, standard stores, no proprietary runtime | Deep dependency on a single vendor’s stack |
Open data connectors live out of the box
Of claims carry source-level provenance
From contract to deployed on your infrastructure
Vendor lock-in — your data in open, standard stores
MCP tools for structured, machine-speed ingest
Autonomous agents advance lines of inquiry, on a leash
Built-in data sources
Magen ships with keyless access to dozens of open feeds, plus keyed connectors for commercial sources — all correlated automatically across time, geography, and named actors.
Secure and governed by design
Database-level isolation
Every tenant table runs under forced row-level security; the app connects as a least-privilege, non-bypass role.
Policy-gated actions
Consequential operations pass an OPA/Rego decision before they commit — no exceptions, no back doors.
Append-only audit
Every mutation lands in an immutable audit log: who, what, when, and the exact change.
Sensitivity & classification
Per-user sensitivity ceilings and enforceable classification banners gate what each analyst can see and do.
Lawful by design
Agents propose, humans approve. No autonomous high-impact decisions, with hard spend rails and a kill switch on every agent.
Sourced or it doesn’t ship
Reports compile only from existing claims, evidence, and sources — every judgment carries its references.
Deploys where the mission lives
Managed cloud
Fully managed multi-tenant SaaS, hardened and isolated per workspace.
On-premise
Run it in your own Kubernetes, on your own infrastructure, under your control.
Air-gapped
Standard Docker delivery for disconnected and classified environments.
Sourced. Defensible. Audited.
Magen gives your analysts a decision advantage they can defend — to a commander, a court, or a coalition partner. Gotham-grade depth, without the black box.
Request a briefing