Resources - White paper

Mobula White Paper

Mobula is CyRay's AI-orchestrated Security Operations Platform. This paper sets out the problem it answers, the architecture that carries its guarantees, how it deploys from cloud to fully air-gapped, and the numbers behind it.

Document 1.0.122 August 2026For enterprise security buyersAll claims approved and sourced

ABSTRACT

Security operations centres do not fail for lack of tooling. They fail because the work that makes a SIEM actually detect things - writing rules, mapping fields, tuning out noise, keeping content current as the estate changes - is continuous, specialist, and almost always understaffed. Underneath every recurring failure is one structural question: how much of security operations can be made deterministic, guaranteed by code and verifiable state, rather than resting on individual heroics and manual supervision.

Mobula is built as an answer to that question. Detection content is engineered centrally, by experts and machines, quality-gated and mapped to MITRE ATT&CK, then packaged, versioned and distributed automatically to every customer environment - and kept current as the threat landscape and the customer's own telemetry evolve. Customers get continuously improving protection without ever writing or tuning a rule.

The paper describes the hub-and-spoke architecture that closes the loop from collection to tuning, the data boundaries verified in code rather than by policy, the collection semantics that make a silent collector impossible by construction, and the Standalone Law under which the entire stack - including the reasoning AI - runs disconnected. It closes with the figures behind the platform, every one measured or drawn from the operating record, and deliberately publishes no estimated or extrapolated numbers.

CONTENTSWhat the document covers

1

The problem in modern SOC operations

Four failure patterns that recur in practice: the rule-authoring burden landing on the customer, collection that silently degrades, parsing that fragments per collector and per site, and air-gapped environments left with a frozen product.

2

What Mobula is

An AI-orchestrated Security Operations Platform in which detection content is engineered centrally by experts and machines, and delivered continuously to every customer environment. Customers never write or tune a rule.

3

Architecture overview

Hub and spoke, closed loop. Versioned, attributed content with one canonical write path. Constitutional data boundaries verified in code. Collection that cannot fail silently. Built for scale.

4

Module capabilities

Nemo, the autonomous detection-engineering agent. HASH Ocean, knowing what runs on the estate. mobula-ai, the reasoning layer that works disconnected.

5

Deployment models

Cloud, on-premises, and fully air-gapped under the Standalone Law: the entire stack, including the AI layer, runs with no outbound connectivity.

6

Differentiators and Mobula in numbers

Eight structural differentiators, and the figures behind them - every one measured from stored platform state or drawn from the operating record, with no estimated or extrapolated numbers.

7

Conclusion and next step

Why security operations improve when the scarce work is engineered once, centrally, under machine-verified discipline, and what a technical briefing covers.

FIGURESMobula in numbers

As printed in the paper. Every figure is either measured from the platform's own stored state or drawn from CyRay's operating record.

100K+ EPS
sustained ingestion
9 years
company track record
5 years
Mobula in production
20+
live deployments
6,000+
expert-engineered detection rules in production
7,500+
MITRE ATT&CK technique mappings across the rule library
4,400+
Sigma rules in the managed detection-content chain
2,300+
connector parser definitions - one field authority per log source
12
integrated modules across the ecosystem

CyRay, the company behind Mobula, holds ISO 27001 certification and a national CERT accreditation. Mobula deployments span eight sectors; consistent with the data boundaries the paper describes, it names no customers. Named references are available on request, under NDA.