Cognition × Knowledge

Any question.
Any data.
Any complexity.

COGknows consolidates every source you hold — documents, databases, open source, field reports, sensors and messaging — into one governed knowledge layer, then answers questions across all of it. Every answer arrives with a RIVER™ trust score, and every trust score comes with a reason — AI answers you can defend.

5
Trust dimensions
12+
Source types
100%
Tenant isolation
ANSWER · grounded RIVER™ 0.82
“Which vendors funded entity Delta in 2025, and how reliable is that?”
Three vendors funded Delta in 2025, corroborated across 4 independent sources. One transfer is flagged near-expiry (license lapsed).
R · reliability0.91 I · integrity0.86 V · validity0.79 E · expiration0.62 Rel · relevancy0.88
Documents/Relational DBs/Open source/Field reports/Sensors/WhatsApp/Telegram/Web scrape/Audio

Two answers. Identical confidence.
Only one deserves it.

To a plain AI assistant, these are the same sentence. COGknows is how your team tells them apart.

“The shipment arrived at the port on June 12.”
Built on an official port record, confirmed by 3 independent sources.
88
Highly reliable
“The shipment arrived at the port on June 12.”
Built on one anonymous forward, relayed through three channels.
21
Questionable
Consolidation

Vast data integration.
One consolidated brain.

Your knowledge is scattered across silos that were never designed to talk. COGknows ingests all of it through a single pipeline — detection, conversion, chunking, embedding, entity extraction — and resolves it into a unified knowledge graph. Ask one question; get an answer spanning every source, at any complexity.

{{ s.label }}
Knowledge Graph
entities · relations · RIVER™
Hybrid retrieval

Full-text, vector and graph search fused — semantic recall with structural precision.

Entity resolution

Duplicates consolidated with an audit trail. One person, one node — across every source.

Automatic ontology

Types and relationships inferred as data arrives — or define your schema by hand.

Connectors

Connect anything you already have.

From production databases to field WhatsApp groups — 30+ connectors feed one governed knowledge layer. Most never copy your data: they map it, describe it, and query it in place.

{{ g.name }}

{{ g.feature }}

{{ c }}
Every external source gets a RIVER™ trust baseline from an 8-question setup wizard — so reliability travels with the data from the moment it connects.
Trust

Not all data is equal.
RIVER™ makes that visible.

Most systems treat every retrieved fact as equally true — and quietly amplify rumours, stale charts and single-source claims. RIVER™ scores every datum on five orthogonal dimensions, propagates the signal from source to chunk to entity to answer, and shows it on every result.

R·I·V·E·R — Reliability · Integrity · Validity · Expiration · Relevancy
R I V E Rel
{{ d.letter }} {{ d.score }}
{{ d.name }}

{{ d.blurb }}

FLAGS: CORROBORATED FIRST_PERSON NEAR_EXPIRY CONTRADICTION SINGLE_SOURCE HEARSAY
A plain AI assistant
Where did the answer come from? Unknown.
How confident should you be? Judge the tone.
Invented details read exactly like facts.
Its knowledge froze at training time.
An expert who knows better has no say.
COGknows with RIVER™
Named sources — each with its own trust badge.
A 0–100 score with a five-part breakdown.
Unsupported claims cut the score and raise an UNGROUNDED warning.
Scores age in real time; stale evidence is flagged EXPIRED.
One-click analyst overrides — on the record, never overwritten.
Solutions

Four ways to act on what you know.

One knowledge layer, four surfaces — build views, ask questions, generate working apps, or see it all on a map. All share the same governed data and the same RIVER™ trust signal.

DASHBOARD BUILDER

Describe a view. Get a dashboard.

Natural-language widgets — charts, tickers, distinct-value panels and live queries — composed on a grid, published with a click, and shared via revocable public links. Every figure traces back to its sources and its trust score.

NL → chart Public share links Embeddable tiles
incident-overview · published
INCIDENTS / WEEK
OPEN CASES
147
▲ 12% wk
AVG RIVER™
0.74
Summarise funding links for entity Delta and flag anything contradicted.
CK
Delta received transfers from three vendors. Two are corroborated across independent filings; one is contradicted by an official registry.
⌖ filing_2025_03.pdf · 0.88 ⌖ registry.gov · 0.91
RICH CHAT INTERFACE

Ask anything. Grounded, cited, scored.

A streaming RAG agent with tradecraft built in — hybrid search, graph traversal, timelines and dossier tools across a multi-step loop. It triangulates, separates fact from inference, flags collection gaps, and grounds every claim against its retrieved context.

17 agent tools Faithfulness check Inline citations
VIBE-CODING APP GENERATOR

Generate apps over your data.

Describe a tool; COGknows writes JS or Python that runs in a locked sandbox — display and calculation only, no OS, no network, reading only whitelisted data through a signed bridge. Embed it as a dashboard tile or share it as a read-only link.

Sandboxed iframe HMAC code-signing JS + Python (WASM)
risk-map.py · signed ✓
df = rashbi.data
  .query("incidents")
risk = score(df)
chart(risk, kind=
  "heatmap")
incident-geo · live kepler.gl · GPU
24
31.78°N · 35.21°E
LAYERS
Pointsagent
Heatmapgraph
Arcsroutes
Clustersincident
Areasimport
GIS MAP VIEW

See it on the map.

A GPU-accelerated geospatial view layers incidents, entities, routes and territories on one interactive map — points, heatmaps, arcs, clusters, lines and polygons. Layers are fed live by the chat agent, the knowledge graph, WhatsApp and your own imports, and every feature keeps its RIVER™ trust signal.

7 layer types Draw & filter Agent-generated layers
Use cases

The data exists. The insight is what’s late.

Wherever answers are buried across disconnected sources, COGknows consolidates them and returns a defensible answer. A few of the questions teams bring us — road safety is one example, not the whole story.

{{ u.tag }}
{{ u.title }}
{{ u.q }}

{{ u.body }}

{{ c }}
Security & Privacy

Isolation is the default, not an upgrade.

For government and commercial intelligence alike, the data is the risk. COGknows is built so that one customer can never see another's — and so that your data is never used to train anyone's model, including ours.

{{ x.tag }}
{{ x.title }}

{{ x.body }}

Per-workspace isolation, all the way down

Each workspace gets its own database, its own graph, its own object store and its own search index. Customer-to-customer isolation is structural — there is no shared query path to leak across.

Mongo db Neo4j db MinIO bucket Search index
Pricing

Priced on what you actually use.

Plans scale with documents, workspaces, external sources and monthly LLM tokens — and the model tier you pick. Pick a starting point, then fine-tune everything in the configurator below.

Indicative pricing · USD and ILS (₪) · FX ≈ 3.70 · billed monthly
POPULAR
{{ t.name }}
{{ t.tagline }}
{{ t.priceUsd }} {{ t.per }}
{{ t.priceIls }}
{{ t.cta.label }}
{{ f }}
Onboarding configurator

Build your plan in four steps.

Choose how smart your models should be, set your scale, switch on the modules you need — the price updates live.

How smart should the models be?
Higher intelligence costs more per token. You can mix tiers per workspace later.
Set your scale
{{ sl.label }} {{ sl.display }}
{{ sl.minLabel }}{{ sl.maxLabel }}
Add optional modules
Switch on only what you need. Manual ontology is included free.
Your configuration
{{ li.label }}
{{ li.detail }}
{{ li.usd }}
{{ recoNote }}
Estimated monthly
{{ totalUsd }} / mo
{{ totalIls }}
RECOMMENDED PLAN
{{ recoTier }}
{{ recoSub }}
Model tier{{ chipModel }}
Monthly tokens{{ chipTokens }}
Documents{{ chipDocs }}
Workspaces{{ chipWs }}
External sources{{ chipSrc }}
Modules{{ chipMods }}
Start with this plan →
Contact

Let's map your sources.

Tell us about your data and your deployment constraints. We'll come back with a scoped plan — managed cloud, private cloud (VPC) or fully on-prem / air-gapped.

Response within one business day
Security review & reference architecture on request
NDA-friendly · your data never leaves the call
Message received.

Thanks {{ sentName }} — we'll be in touch within one business day.

No spam · used only to respond to your enquiry