Report in preparation  ·  First edition 2026  ·  Deployments accumulating
// Annual Benchmark  ·  Edition 01

The State
of Operational
Predictability.

The first annual benchmark on operational learnability across high-risk environments. How consistently operations behave. How quickly adversaries build working models. What the patterns look like across sectors, corridors, and threat classifications.

Status In preparation
Edition 01  ·  2026
Coverage Mining · Infrastructure · Logistics
Data source Zerathis Blindspot deployments
Client names Not disclosed
// What this report is

Every Zerathis Blindspot deployment produces a Blindspot Audit — a 30-day intelligence assessment of what an adversary can observe, model, and act on at a specific site. Each audit produces a learnability score, a MAP stage classification, an exposure window profile, and a set of insider and environmental risk indicators.

Across multiple deployments, these assessments produce something more valuable than individual site data: patterns.

How learnable are sites before intervention, on average? What are the most common sources of predictability across sectors? How quickly do adversaries move from initial observation to execution confidence? What reconnaissance indicators appear most consistently before a high-confidence exploitation attempt?

No competitor has this data. It exists only in systems that have been deployed long enough, across enough environments, to see patterns rather than incidents.

The State of Operational Predictability will be published annually by the Parthenius Air Intelligence Unit. Client identity is never disclosed. All data is aggregated and anonymised. The patterns are the product.

// Report contents  ·  Edition 01
Publication date to be confirmed  ·  2026
01
Learnability Score Distribution
Median, range, and sector breakdown of site learnability scores at audit entry across all Zerathis deployments
Pending
02
Common Sources of Predictability
The operational patterns most frequently identified as adversarial learning vectors: shift timing, patrol consistency, weather response, infrastructure access
Pending
03
Reconnaissance Indicator Patterns
Probe frequency profiles, approach vector clustering, and timing signatures that precede high-confidence exploitation attempts
Pending
04
MAP Stage at First Detection
Distribution of adversarial planning stage when sites first engage Zerathis — how far along the model-building cycle had already progressed
Pending
05
Environmental Exploitation Patterns
Weather, shift transition, and maintenance window exploitation across sectors — how adversaries use operational variables as planning tools
Pending
06
Insider Risk Signal Profiles
Categories and frequencies of insider-enabled intelligence identified across deployments, without individual attribution
Pending
07
ODR Trajectory by Threat Classification
How Opportunity Denied Rate develops over deployment time across battlefield and contained-environment classifications
Data accumulating
08
Adversarial Adaptation Profiles
How organised networks respond to deterrence — fragmentation patterns, timing shifts, and counter-surveillance evolution observed across active deployments
Data accumulating
// How the data is collected
01  ·  Source
Live Deployments Only
All data originates from active Zerathis Blindspot deployments — sites where the full five-stream intelligence architecture is running continuously. No survey data. No self-reported metrics. No modelled estimates. Everything in this report is drawn from the operational record.
02  ·  Anonymisation
Pattern Without Identity
Client identity is never disclosed. Site names, geographic specifics, and any identifying operational detail are removed at the point of aggregation. What remains is the pattern — the learnability score, the MAP stage, the probe frequency, the ODR trajectory. The pattern is the product.
03  ·  Validation
Independent Review
All aggregated metrics are reviewed by a qualified analyst before publication. Patterns are validated against the raw deployment record. No finding is published without sufficient deployment volume to be statistically meaningful. Where data is insufficient, the section is withheld until the next edition.
// What we already know

The first edition of this report will draw on deployment data that is already accumulating. Two findings from current active deployments give early indication of what the benchmark will show.

These are not projections. They are operational outcomes from the first Zerathis deployment, published here as the prior data that the full benchmark will contextualise across multiple sites.

Platinum mining  ·  Limpopo corridor  ·  Month 18
97%
Opportunity
Denied Rate
8.7/10
Learnability Score
at audit entry
MAP classification: Extreme  ·  Insider risk: Confirmed  ·  Weather exploitation: Confirmed
Adversarial adaptation: Large organised groups → professional fragmented cells
Silence period: 3 months continuous at month 12
Critical infrastructure  ·  Freight rail  ·  Gauteng
70%
Incident
Reduction
Pre-
Zerathis
Metric captured before
ODR framework deployed
Methodology origin deployment  ·  Division-of-labour syndicates  ·  Decoy tactics confirmed
Environmental timing: Adverse weather used as operational variable
This deployment is why we stopped counting incidents
// Include your deployment

Your deployment
data contributes
to the benchmark.

Operators who commission a Blindspot Audit or deploy Zerathis Blindspot are invited to include their anonymised deployment data in the annual benchmark. Participation is optional and requires written consent.

Participating operators receive the full benchmark report in advance of public release, including the sector-specific data relevant to their deployment environment. They also receive a personalised benchmarking addendum comparing their site’s metrics to the anonymised sector average.

No identifying information is included in any published output. The contribution is the pattern. The return is context.

// Participation requires an active Zerathis Blindspot deployment or completed Blindspot Audit. Consent documentation provided on request.
// Express participation interest

// No commitment required to express interest.
// Treated as strictly confidential.
Noted.
We will be in contact as the first edition takes shape.
Your data contributes to a benchmark nobody else is building.
// First edition notification
Know when
it publishes.

The first edition will be distributed directly to registered recipients before public release. No subscription. No marketing. One notification when the report is ready.

// One email when the report publishes. Nothing else.
Noted  ·  You will be among the first to receive it.