IAM RAIN

A Defence AI OS, not a product you install

An operating system is judged by what runs on top of it. Ours carries data from any sensor to any command system, runs reasoning on that data in place and keeps the picture current whether or not an exercise is happening.

Why an OS and not a platform tool

Military AI capability cannot be procured off the shelf. It has to be grown inside a force through continuous operation. Capability that only appears during an exercise decays the moment the exercise ends.

IAM RAIN owns the corridor through which external capability reaches NATO C2. That is an architectural position, not a product feature. It makes us a load bearing part of the operating picture rather than one more supplier in a queue.

Continuous operation between exercises is what turns a demonstration into a standing capability. That is the part of the work that does not show up in a demo.

Architecture overview

Defence AI OS architecture, four horizontal bands Four stacked horizontal bands, top to bottom: Edge, Fabric, Reasoning, Output. A fifth band beneath them, Identity and access, cuts across all four. Edge Sensors, UAVs, vehicles, IoT, 5G/NTN links Fabric Streaming, storage, sovereign object store Reasoning Confidential inference, RAG, agent orchestration, vector search Output C2, CRC, CAOC, TAK/ATAK, WebTAK Cross cutting: identity and access, role based access, Kubernetes native, platform console
Text equivalent

Four horizontal bands, top to bottom. Edge: sensors, UAVs, vehicles, IoT, 5G and NTN links. Fabric: streaming, storage and sovereign object store. Reasoning: confidential inference, retrieval-augmented grounding, agent orchestration and vector search. Output: C2, CRC, CAOC, TAK/ATAK, WebTAK. A fifth band cuts across all four: identity and access, role based access, Kubernetes native deployment and the platform console.

The three pillars

Data Fabric

Ingress standards, streaming, storage and the connector library that carries sensor data into the platform.

Agentic AI Layer

Orchestration, retrieval-augmented grounding and confidential inference that turn raw data into operationally relevant output.

Digital Nervous System

Real time awareness, multi-site operation and degraded-link behaviour that keep the picture live.

What this replaces

The usual route to reaching a command system, compared with the Defence AI OS
The usual route With the Defence AI OS
Bespoke point to point connections per vendor One fabric, one connector library
Capability appears for the exercise, then stops Continuous operation between exercises
Data leaves sovereign control Sovereign compute, air gapped deployable
Programme timelines measured in years New capability reaching C2 in days

Book a technical briefing