Navalia | Material-change intelligence engine | Cognous
Navalia · Material-change intelligence · Working software

What materially changed — and how do we know it is true?

Navalia is a material-change intelligence engine. It detects verifiable change in the world, filters non-material noise before inference, evaluates claims through a procedural trust chain, clusters overlapping evidence into canonical nodes, and publishes a structured daily map of what actually changed.

The platform Request a walkthrough
8 operations
signal to canonical node
Filter first
razors run before inference
Daily
structured intelligence map
What it is

Eight operations, in order.

A material change is an event that significantly alters circumstances, policies, or perceptions. Navalia is the system that finds those events and establishes whether they hold.

01
Detects

Verifiable changes in the real world, from wires, agencies, briefings, and primary filings.

02
Filters

All non-material noise — removed by deterministic razors before any model sees it.

03
Normalizes

Heterogeneous inputs into a single unified schema.

04
Evaluates

Claims through a procedural trust chain rather than a source-reputation score.

05
Clusters

Overlapping evidence into canonical nodes, collapsing duplication.

06
Scores

Importance from primary-source density and consensus.

07
Synthesizes

Context, timelines, and the actors involved.

08
Publishes

A daily, structured map of what actually changed.

This is not news, summarization, search, research automation, or feeds. It is intelligence infrastructure for the civilian world.

Why it is a new category

Every existing category optimizes for something Navalia rejects.

Existing category
Optimization target
Why it fails
News
Engagement
Rewards noise, speed, and outrage
Aggregators
Volume
Redundant, unverified, repetitive
Search
Relevance
Requires user intent; no synthesis
Research tools
Retrieval
No trust chain, no clustering
LLM summaries
Compression
Hallucinates; no evidence model
Feeds
Personalization
The opposite of structural understanding
Navalia
Structural understanding of material change
The target no current category addresses
The pipeline

Six stages from raw signal to canonical node.

Filtering happens before inference, not after. By the time a model participates, the input is deduplicated, language-checked, and heuristically cleaned.

I
01
Raw signal ingestion

Wires, agencies, briefings, filings, and early reports enter as a deliberately chaotic, overlapping set — the real shape of the world.

II
02
Razor filters

Deduplication, language consistency checks, and rule-based heuristics remove noise before any model is invoked.

III
03
Claim extraction

Information is broken into atomic, neutral, evidence-linked claims, each timestamped and weighted by source reliability.

IV
04
Procedural trust chain

Source quality, then verification, then synthesis. Trust is a procedure that can be inspected, not a score that must be believed.

V
05
Clustering

Related claims collapse into canonical nodes representing the actual development, rather than the many reports about it.

VI
06
Bioptic evaluation

Micro-truth and macro-pattern analysis are read together to set importance and rank what genuinely matters.

Output

A daily briefing that carries its own evidence.

◎
Evidence-linked canonical nodes

Every node in the daily map resolves to the claims and primary sources that support it. Nothing asks to be taken on faith.

≡
Contextual scaffolding

Each development arrives with timeline, actors, and prior state.

―
Tone-neutral presentation

No engagement framing, no urgency signalling. Neutral tone is a design constraint, not a style preference.

⊞
Structural comparability

Because outputs share one schema, today can be compared to yesterday — and drift in the record itself becomes visible.

Stress test

The hardest case is the first 24 hours.

Navalia is demonstrated against historical high-chaos events, where reporting was contradictory, sources were uneven, and the record only settled later. The point of the exercise is not retrospective judgement — it is to show how the pipeline behaves when the input is at its worst.

Without structure
Conflicting reports propagate unchecked Volume is mistaken for confirmation Drift and contradiction accumulate Decisions rest on unverified input
With Navalia
Claims are atomic and separately verifiable Evidence density drives importance Canonical nodes replace duplicate reports Uncertainty is stated rather than smoothed
Where it runs

Navalia is an application of the platform, not a separate stack.

01
On the ISS substrate

Continuity and constraint gating keep synthesis anchored to prior state across long analytical runs.

ISS →
02
Under the control plane

Authority, admissibility, and certification states govern what the engine is permitted to publish.

Platform →
03
With portable evidence

Runs can be packaged and evidenced using the open artifacts published in Open Control Stack.

Open Control Stack →

See a day of material change, structured.

Navalia is working software. A walkthrough runs against real signal.

Talk to Cognous See the platform