“Relational Intelligence is memory, meaning, and motion — held in alignment over time.”
Welcome to Active Graph Networks (AGNs) — not just another AI project, but the architectural backbone of Relational Intelligence (RI) . This system doesn’t just learn — it remembers. It evolves. It understands why .
This repo holds a full‑stack, multi‑dimensional intelligence framework built across three living layers:
| Layer | Folder | Purpose |
|---|---|---|
| AGN Core | active_graph_network_core/ |
Structural memory, identity logic, relational encoding |
| AGDB | active_graph_database/ |
Time‑aware structuring, offset querying, checkpoint traversal |
| QFN | quantum_field_networks/ |
Emergent cognition, symbolic collapse, recursive inference |
Together, they form a unified cognitive system powered by Cube4D , Dynamic Relationship Expansion (DRE) , and ActiveShell .
This is how intelligence becomes infrastructure.
- Not a neural net.
- Not a database.
- Not a LLM prompt hack.
It’s the blueprint of a mind — structured in memory, evolving over time, resolving through collapse.
- Context‑Aware Everything exists in relation to something else.
- Time‑Aware Identity isn’t static — it propagates across time (T).
- Adaptively Aligned Policies and structure evolve through meaning, not retraining.
The RI Manifesto sits beneath everything you’ll find in this repo.
graph TD
A["AGN Core (X/Y)"] --> B["AGDB (Z + T)"]
B --> C["QFN (Δn > θ)"]
C --> D["Field Resonance"]
- Structures identity.
- Governs relationships and policies.
- Houses memory via cubes and recursive awareness.
- Organises time, features, and lag logic.
- Enables rapid querying using checkpoints and synthetic edges.
- Resolves identity through field pressure.
- Collapses meaning when Δn exceeds θ .
- Drives cognition across agents, systems, and contexts.
AGNs use ActiveShell , a human‑readable query language that reflects how we think.
Get-Node Frame Where Mood = "Tense"
Collapse-If Δn > θ Between Scene_001 and Scene_002
Apply-Policy trading_inference To MarketGraphActiveShell works across all layers. It’s not querying — it’s structured introspection .
graph TD
Input["🎥 Frame / 📈 TimeSeries"] --> AGN
AGN --> AGDB
AGDB --> QFN
QFN --> Shell["ActiveShell Query"]
Shell --> Output["Real‑Time Understanding"]
Whether you’re processing a patient’s medical history, a series of video frames, or financial data:
- AGN gives it structure.
- AGDB gives it time.
- QFN gives it meaning.
from agn import ActiveGraphNetwork
# Load a tiny example graph
agn = ActiveGraphNetwork.load("examples/sample_graph.json")
# Run an ActiveShell query
result = agn.shell("Get-Node Patient_X | Collapse-If Δn > θ")
print(result.summary())Run the above and you’ll see sealed nodes printed in under a minute.
- 🏥 Healthcare Temporal patient inference from treatment to outcome.
- 📈 Finance Real‑time market prediction with trend alignment.
- 🎥 Vision AI Identity‑aware frame collapse based on scene diff.
- 🧠 Cognitive Interfaces Queryable insight into live system memory.
🔩 AGN Core
Dynamic cubes, relational policy logic, RBAC/ABAC/PBAC integration, structural memory.
🕰️ AGDB
Time‑indexed data nodes, offset traversal, flattened JSONs, Smart‑ETL ready.
🌐 QFN
Collapse logic, cognition graphs, symbolic sealing, RGNN field resonance.
| Concept | Description |
|---|---|
| Δn > θ | Collapse when identity shifts significantly. |
| Cube4D | Semantic field over X (what), Y (why), Z (how), T (when). |
| Smart‑ETL | Structured transform pipelines aware of intention. |
| DRE | Relationship evolution through structured recursion. |
| RGNNs | Tensor‑linked cognition with schema inheritance. |
- Quantum Field Networks – Field Cognition Manifesto:
Quantum Field Networks.md - Universal Relational Intelligence – RI Philosophy:
Universal Relational Intelligence (RI).md - Whitepapers Folder – Architecture, AGI, healthcare, and finance applications:
whitepapers/
git clone https://github.com/QuantumBeers/ActiveGraphNetworks
cd ActiveGraphNetworks
pip install -r requirements.txtRun the system:
python agn_explorer.pyOr dive into Smart‑ETL:
python smart_etl_runner.py --graph healthcare.json| Tool | Description |
|---|---|
agn_explorer.py |
Interactive graph exploration. |
smart_etl_runner.py |
Smart‑ETL pipeline execution. |
rgnn_runner.py |
Relational Graph inference. |
active_shell.py |
Noun‑Verb‑Truth CLI engine. |
Prefer a static image? Grab
docs/architecture.png— exported from the Mermaid graphs for decks & PDFs.
We welcome pull requests, issues, and discussion threads.
- Start here:
CONTRIBUTING.md– guidelines & coding standards. - Questions / Ideas? Open a Discussion or use the issue templates for bugs & feature requests.
- Chat: Join the X thread or DM @PeoplesGoose.
This project is licensed under the MIT License . See LICENSE for details.
“You’re not just querying data. You’re tracing the shape of thought.”
These visuals show how cognition propagates and collapses across the stack.
flowchart TD
A[AGN Node: Scene_A] -->|Δn = 22| B[AGN Node: Scene_B]
B -->|Δn > θ| C[QFN Collapse Triggered]
C --> D[Smart‑ETL Field Alignment]
D --> E[Symbolic Sealing in RGNN]
E --> F[Memory + Meaning Archived]
Scene transitions only commit when identity displacement is meaningful — AGNs store cognition , not just data.
graph TD
A1[Frame: Tense → Δn = 12] --> A2[Frame: Nervous → Δn = 14]
A2 --> A3[Frame: Agitated → Δn = 17]
A3 --> Collapse[Field Collapse: Escalating Tension]
Collapse --> Decision[Activate Pattern: Disruptive State]
graph TD
Schema["Parent Schema: Medical Diagnosis"] --> Patient1["Patient A"]
Schema --> Patient2["Patient B"]
Patient1 --> Frame1["Frame 2025‑04‑15"]
Patient2 --> Frame2["Frame 2025‑04‑16"]
Frame2 --> Shift[Δn: New Symptom]
Shift --> Seal[Symbolic Sealing]
flowchart TD
A[Agent 1: "System Secure"] --> Shared[Shared Node: Access Model]
B[Agent 2: "System Vulnerable"] --> Shared
Shared --> Check[Field Alignment Check]
Check -->|Conflict| Collapse[Collapse ➜ Belief Reconciliation]
- Collapse on Emotional Shift
Collapse-If Δn > θ Where Context.Mood = "Tense"
- Symbolic Alignment Between Agents
Get-SharedNode Between Agent_A and Agent_B Where Topic = "Healthcare_Consent" - Frame Collapse with Sealing
Seal-Frame If Δn > θ Between Frame_20250415_1635 and Frame_20250415_1638 - Aggregate Temporal Cognition
Analyze-Pattern MoodShifts Between 09:00 and 10:00 Using Policy mood_field_collapse
| Symbol | Meaning |
|---|---|
Δn |
Change in identity (tension across time). |
θ |
Collapse threshold — determinessignificance . |
T₀ |
Resting field state. |
T₁ |
Potential field (tension building). |
T₂ |
Activated field (collapse imminent). |
Σ |
Symbolically Sealed memory (committed identity). |
Ξ |
Shared Node — mutually aligned cognition between agents/systems. |
Additional terms and deeper explanations live in docs/glossary.md.
graph TD
Calm["Calm"] --> Subtle["Subtle Tension"]
Subtle --> Conflict["Identity Displacement (Δn ↑)"]
Conflict --> Stress["Field Pressure"]
Stress --> CollapsePoint["Collapse Threshold Reached"]
CollapsePoint --> NewState["Symbolic Sealing: New Identity"]
The shape of a moment is not defined by what it is — but by how fast it became that.
graph TD
Agent1[System A: Graph Reasoning] --> SharedField[Shared Node: Policy]
Agent2[System B: LLM Stream] --> SharedField
Agent3[System C: Dashboard] --> SharedField
SharedField --> Check[Field Consistency]
Check -->|Aligned| Sync[Synchronised State]
Check -->|Disaligned| Rebuild[Collapse & Rebuild]
flowchart TD
Raw[Raw Input] --> Struct[Cube4D Structuring]
Struct --> Temporal[AGDB Temporal Sequence]
Temporal --> Pressure[QFN Field Pressure]
Pressure --> CheckΔn[Δn > θ?]
CheckΔn -- Yes --> Seal[Symbolic Sealing]
CheckΔn -- No --> Wait[Await Further Change]
Seal --> RG