Intelligence Operating Network System · v0.4 — Cognitive Attention Architecture

Intelligence from
traversal, not weights

An open protocol for publishing, connecting, and traversing typed knowledge claims. Any model can query it. Any node can join it. The network is the durable asset.

Published CBBs
Typed Relationships
7
Cognitive Domains
Active Nodes
The protocol

Three primitives.
Infinite composition.

IONS externalizes intelligence into a traversable network of atomic knowledge claims. Models are replaceable interpreters. The network is the durable asset.

Cognitive Building Block
A single atomic claim with evidence, confidence, provenance, and a cryptographic hash. Not a document chunk — a typed, addressable knowledge artifact with a vector embedding.
Typed Relationship
Connects CBBs with semantic type — supports, contradicts, causes, depends_on, extends. Relationships are first-class artifacts with confidence, rationale, and cryptographic hash.
Reasoning Path
The stored result of traversal — saved as a reusable artifact with three independent scores: path confidence, path relevance, and path utility from feedback.
🧭
Cognitive Attention
v0.4 routes queries through 7 Cognitive Domains and 24 Subdomains before traversal begins. Attention allocation makes the protocol scale to any network size.
⬡⬡
Federation
Any node implementing the protocol can register and participate. Queries traverse all connected nodes automatically. Node failures never block a response.
🔄
Self-Validating Loop
Routing sessions, path feedback, and structural checks create a learning loop that improves attention allocation over time — no external model dependency.
Query flow · v0.4
Query
+ Intent
🧭
Route
Cognitive Domains
🔍
Discover
Embedding similarity
Beam Search
Guided traversal
📊
Rank
Relevance × Conf × Utility
Synthesize
Grounded answer
📋
Store
Path + Session
Live demo

Ask the genesis network

Queries traverse 8,000+ typed knowledge claims across 7 Cognitive Domains via beam search. Every answer shows its reasoning chain, path confidence, and routing confidence.

How does institutional memory compound competitive advantage?
Why do AI pilots succeed but fail to scale?
What makes knowledge reusable across an organization?
How does proof of work create scarcity in Bitcoin?
What is the relationship between flow states and peak performance?
3 queries per minute · 10 per hour · Genesis node · run your own node for unlimited access
Differentiation

How IONS differs from
RAG and knowledge graphs

RAG
Chunks discarded after query
No typed relationships
No path confidence score
No evidence lineage
No reusable reasoning paths
~ Retrieves relevant context
Knowledge Graph
No path artifact lifecycle
No confidence scoring
No evidence references
No self-improving routing
~ Typed edges
~ Graph traversal
IONS
Reasoning paths stored as artifacts
8 typed relationship types
Three-score path ranking
Cryptographic integrity
Self-validating learning loop
Open protocol, MIT license
Get started

Three ways to participate

01 — QUERY
Use the genesis node
Query 8,000+ CBBs across 7 Cognitive Domains via the demo above or the full explorer. No account required.
Open explorer →
02 — RUN
Run your own node
Clone the repo, add your OpenRouter key, and start your node in minutes. Runs fully private — no external connections required.
git clone github.com/nomad505050/ions-genesis cp .env.example .env docker compose up -d
03 — FEDERATE
Join the network
Register your running node with the genesis node. Your CBBs immediately participate in federated traversal. Optional — run private if you prefer.
curl -X POST \ https://api.ionsprotocol.org/nodes/register \ -d '{"node_id":"your_id", "public_api_base":"https://..."}'