A governed, multi-model AI environment delivered to staff and students within a fixed 10-week academic-cycle timeline.
.png){kind=small}
A governed, multi-model AI environment delivered to staff and students within a fixed 10-week academic-cycle timeline.
A research-intensive UK university - with up to 54,000 staff and students and a planned merger set to increase its student population to approximately 47,000, needed to move from an unmanaged landscape of consumer AI tool use to a formally governed institutional platform. With 92% of UK students using AI tools for their studies and sector-wide academic misconduct costs estimated at £12.4 million annually, the institution required an enterprise-grade solution within a fixed academic-calendar window.
The university had no role-based feature controls, no acceptable use policy enforcement, no usage analytics, and no safeguarding infrastructure. Staff and students had no sanctioned platform and no formal guidance on appropriate AI use.
With no governed platform, the university could not restrict AI answer generation during assessments, distinguish AI-assisted from AI-generated work, or produce audit trails for misconduct investigations. AI detection tools achieve only 33-81% accuracy, an unreliable foundation for academic integrity policy.
Deploying AI to tens of thousands of users - including under-18 students, required a formally executed Data Processing Agreement, a DPIA, KCSIE 2025-aligned safeguarding controls, and automated data subject rights workflows before any student personal data could be processed.
Staff access was required by March 2026 and student access by April 2026 - a 10-week implementation window. Student launch was deliberately post-Easter to avoid the Spring assessment period. These dates were fixed and feasibility had to be proven before contract award.
The planned merger would increase the combined student population to approximately 47,000. The platform had to be validated at that scale before deployment, with a commercial model accommodating the enlarged user base without full renegotiation.
PromptX separates concerns across four independent layers: a progressive web application across all devices; a governance layer with role-based access control across eight configurable dimensions; a multi-model orchestration layer routing across Claude Sonnet 4.5, GPT-4 Turbo, and Gemini 2.0 with automatic failover; and a differential privacy analytics layer (epsilon=1.0) providing institutional insight without individual identification.
VE3 configured tailored profiles for each group: Academic Staff (full access - assistants, all models, file uploads, API credentials); Professional Services (administrative tools, policy support); Postgraduate Researchers (code and data workflows, API access); Undergraduate Students (safeguarded access, Learning Mode, assessment restrictions); and Postgraduate Taught Students (matched profile with appropriate extensions).
SAML 2.0 integration with the university's Microsoft Azure AD / Entra ID, SCIM 2.0 for automated user lifecycle management, and domain claim via DNS TXT record enabled migration of existing consumer accounts, MFA enforcement through the university's existing authenticator, and full institutional control from a single administrative plane.
Learning Mode refuses direct answers to assessment questions, guiding students through Socratic reasoning instead. Scheduled configuration automates assessment period lockdowns - restricting file uploads and advanced models during examination weeks and restoring access during coursework - without manual intervention. Tamper-evident WORM audit logs are retained for 7 years to support any misconduct investigation.
Phase 1 Mobilisation (Wks 1–2); Phase 2 Identity & Domain Claim (Wks 2–3, parallel); Phase 3 Administrative Configuration (Wk 4); Phase 4 Staff Pilot - 500–1,000 digital champions with defined success criteria (Wks 5–6); Phase 5 Staff Launch - 4,000–8,000 accounts, 16 March 2026 (Wk 10); Phase 6 Student Launch - 17,000–46,000 students, 13 April 2026 (Wk 14).
.png)
By running identity integration, governance, compliance, and training as parallel workstreams from day one, VE3 delivered a fully governed institutional AI environment on schedule, within budget, and meeting 100% of mandatory requirements through native configuration. The university moved from zero formal AI governance to a compliant, multi-model AI platform in 10 weeks - with academic integrity, safeguarding, and data protection built in from the start. "PromptX is a production-ready, enterprise-grade platform specifically designed for institution-wide deployment in UK higher education - meeting 100% of stated mandatory requirements through native configuration rather than custom development."
.png)
