Skip to content
GEA

How cognitive sovereignty decomposes into replaceable parts

The eight-layer architecture

The architecture treats AI-mediated learning as a stack — eight interoperating layers, each replaceable, each with its own governance. L1 is the only actor; everything else is a resource, a principle, or a constraint. L2 is the pivot — every other layer either feeds it or is constrained by it. Remove any single layer and the system collapses.

The eight-layer architectureL1 (the Mentor) is the sole actor, taking three inputs: knowledge structure from L5, learner understanding from L2, pedagogical principles from L3. L2 ↔ L3 is bidirectional. The depth of L2 forces L4 into existence. Evidence (L6) is anchored both to the profile (L2) and to the graph (L5). L8 regulates the entire stack; L7 carries it physically.L1The MentorL2The Learner ProfileL3The Didactic ModelL4Data SovereigntyL5The FederatedKnowledge GraphL6The CompetencyEvidence SystemL7Autonomous Hub InfrastructureL8Trustee Governance
L1 (the Mentor) is the sole actor, taking three inputs: knowledge structure from L5, learner understanding from L2, pedagogical principles from L3. L2 ↔ L3 is bidirectional. The depth of L2 forces L4 into existence. Evidence (L6) is anchored both to the profile (L2) and to the graph (L5). L8 regulates the entire stack; L7 carries it physically.
  • L1

    The Mentor

    The system's only actor

    Not a chatbot that answers questions — a counterpart that knows the learner and accompanies them through explanation and silence, encouragement and provocation. An ensemble of six specialised agents (Content, Didactics, Navigation, Perception, Protection, Audit) operates on different time horizons. Three tiers of generation: real-time Socratic dialogue, AI-generated simulation environments, and long-lived deterministic engines that compute the actual physics.

    Read the layer
  • L2

    The Learner Profile

    The pivot

    A compression pyramid: from raw clicks and hesitations through real-time interpretation and daily synthesis to a long-term profile spanning years. Plus an episodic memory for the moments that matter — "November 12th, cracked vector calculus." No expiration date: the profile accompanies a person from childhood literacy through adult reskilling. Within it lives a personal competency graph projected onto L5.

    Read the layer
  • L3

    The Didactic Model

    The score for L1's performance

    L3 defines the principles by which learning happens; L1 executes them in the moment. Seven principles: intent-based navigation, the Socratic method, dynamic difficulty, functional embedding, content polymorphism (the musician learns fractions as rhythm intervals), progressive de-adaptation that systematically reduces support as competence grows, and collaborative learning that connects learners across hubs and time zones.

    Read the layer
  • L4

    Data Sovereignty

    Violation impossible, not merely forbidden

    Stateless processing combined with stateful encrypted storage under the learner's control — the same architectural pattern Apple's Private Cloud Compute proved at global scale, applied to the most intimate dataset humans accumulate. Custodial keys for children with threshold cryptography and graduated transfer to adulthood. The depth of L2 is what forces L4 to be load-bearing — without it, the deep profile becomes the most dangerous surveillance instrument ever built for education.

    Read the layer
  • L5

    The Federated Knowledge Graph

    The structure of knowledge itself

    A two-layer graph: a universal STEM core globally maintained by independent rotating expert panels, and a sovereignly curated cultural layer for history, society, values, ethics. Models dependencies, not sequences ("fractions require basic arithmetic" yes; "first fractions, then decimals" belongs in L3). Internal layers for concepts, manifestations across domains, competencies, and cross-cutting skills. Becomes irreversible through use — adoption, not regulation.

    Read the layer
  • L6

    The Competency Evidence System

    Competence emerges from process, not exam

    Evidence accumulates from the accompanied learning process, never from separate examinations. Three dimensions: autonomy (did the learner solve it independently?), transfer (can they apply it in a new context?), reflection (can they explain why?). Inherently fraud-resistant: faking months of Socratic dialogue costs more cognitive effort than learning the material. Separation of powers: the Mentor generates evidence; an independent Audit-Agent verifies it; the Knowledge Graph anchors it.

    Read the layer
  • L7

    Autonomous Hub Infrastructure

    Physical access for everyone, everywhere

    A standardised, transportable core module — GPU server, graph database, encrypted storage, simulation engine, satellite uplink, solar power, robust tablets — deliverable by truck, ship or helicopter. Three operating modes: Cloud (full connectivity), Hybrid (heavy tasks remote, core local), Island (degraded but functional, no internet required). Three hub types from Basis (dozens of learners) to Regional (thousands plus physical labs).

    Read the layer
  • L8

    Trustee Governance

    Guardian of the cable, not author of the message

    A Global Education Alliance — analogous to CERN, ICANN or the internet's governance bodies — operates as trustee of the infrastructure: responsible for operation, data sovereignty enforcement, access guarantees, and the universal STEM core. Not responsible for cultural content, didactic decisions or learner evaluation. Whoever operates the infrastructure must not control the knowledge. Emerges in three phases: foundation, consortium, international organisation.

    Read the layer