Preparing your workspace
H2E is a three-layer ecosystem: an Association that governs the research, an Engineering Framework that turns it into reproducible architecture, and Drivia Business Solutions that deploys it as a compliance-ready LMS.
H2E separates research governance, technical architecture, and business deployment into three distinct layers — each with its own mandate and accountability.
The social and philosophical layer. The Association is the living research community that defines the principles, reviews the research, and governs how H2E evolves. Invite-only membership for researchers, builders, and governance leads.
The structural and technical layer. Four subsystems — SROI, NEZ, IGZ, V-RIM — that form a reproducible adaptive intelligence architecture. Formalized in a forthcoming NeurIPS 2026 submission by Wilson Guenther and Boeing Fellow Frank Morales. This layer is the bridge between theory and deployment.
The deployable business layer. Drivia is the reference implementation that turns the H2E architecture into a compliance-ready LMS with clickable citations, mastery gating, 25+ languages, and JAX as the Socratic AI tutor. Built for K-12, higher ed, workforce development, and enterprise.
Every AI response in the Drivia reference implementation passes through these four modules before reaching the human. Together they form the H2E governance loop.
Measures real learning value per interaction — not engagement, not retention. Built on the Ebbinghaus retention curve, calibrated against expert outcomes. The SROI 0.9583 gate ensures every AI interaction clears a measurable learning threshold before the system advances.
Adaptive ZPD. The system continuously chooses to scaffold, challenge, maintain, consolidate, or interleave based on live state — Vygotsky meets Csikszentmihalyi flow.
Classifies the human's real intent across eight categories — explore, verify, vent, draft, decide, learn, contest, off-topic — and governs how the AI responds. Intent governs behavior; behavior governs trust.
Multi-tenant data partitioning so AI can never leak across organizational boundaries. Five isolation layers: Organization → Course → Student → AI Context → Assessment. Compliance-first, by construction.
Co-developed by Wilson Guenther (founder & CTO, Drivia) and Frank Morales (Boeing Fellow). The H2E Engineering Framework is formalized in a forthcoming NeurIPS 2026 submission. Drivia is the reference implementation — every member who joins the Association is standing inside the system the framework runs on.
Self-taught engineer. 1,891+ commits across six products. Built the H2E reference implementation and the Drivia platform from scratch.
Author of the H2E mathematical formalism: Riemannian manifold–based drift detection, SROI gating, and geometric intent classification. NeurIPS 2026 submission under review.
Researchers, builders, governance leads. Tell us who you are and what you'd bring to the room.