The problem is not tool scarcity
Most universities can access AI tools, workshops, and policy templates. What they lack is an explicit architecture for how identity, curriculum, capability, assessment, evidence, decisions, and platforms relate as one institutional system. Without that architecture, AI adoption fragments into pilots that do not compound.
What institutional architecture is
Institutional architecture is the reference model for an AI-native educational institution. It defines the primitives an institution must own, the intelligence loops that must operate across student, faculty, department, and leadership levels, and the boundary between institution-owned capability and platform-enabled services.
- Primitives: the durable building blocks of institutional intelligence
- Relations: how curriculum, capability, assessment, evidence, and decisions connect
- Ownership: what must remain institutional versus what can be platform-enabled
- Verification: how transformation outcomes are observed, claimed, and evidenced
Why architecture before acceleration
Acceleration without architecture produces activity without institutional memory. Faculty experiments remain personal. Student portfolios remain episodic. Leadership dashboards remain vanity metrics. Architecture is what turns implementation into reusable institutional intelligence.
What WG001 will produce
This working group will publish a sequence of working notes, then a reference model, then pilot reports. The goal is not a manifesto. The goal is a transferable architecture that institutions can implement, evidence, and improve.
- WG001-WN001 — Why Institutional Architecture Matters
- WG001-WN002 — Canonical Institutional Primitives
- WG001-WN003 — Capability, Evidence and Decision Architecture
- WG001-FW001 — AI-Native Educational Institution Reference Model
- WG001-PR001 — Pilot Implementation Report
An open research invitation
WG001 is forming in public. Universities, faculty, researchers, industry practitioners, and student researchers are invited to challenge the problem statements, contribute source material, co-author later notes, and host validation pilots. ATRISI convenes the network; the architecture should be co-developed with institutions that will operate it.