Sessions use campus examples to discuss practical tradeoffs, with measured results where available. These topics are starting points for a session shaped around your audience.
Institutional AI: From Pilot to Service
How UC San Diego grew TritonGPT from an early pilot into a supported campus service with 89%+ self-hosted traffic. Covers architecture, governance, on-premises hosting at SDSC, and lessons from real campus adoption.
Audience takeaway: What it takes to move from a pilot to a supported, accountable campus service.
Agentic AI Workflows in Higher Education
How supervised AI agents are moving from personal productivity into core university operations, including contract review, transcript matching, and institutional data Q&A. Covers controls, human review gates, and measured ROI.
Audience takeaway: How to pick workflows with real operational payoff and decide where human review belongs.
Governed Agent Workspaces and Developer Enablement
How UC San Diego equips developers and campus builders with one Gateway key, the TritonAI Harness, reusable skills, and a clear promotion ladder from local prototype to enterprise service.
Audience takeaway: How to help campus teams build with AI without duplicating platforms or losing governance.
AI Governance and Data Foundations in Practice
A practical look at data classification (P1-P4), gateway metering, review gates, and why weak data foundations stall AI programs. Covers where governance has helped and where hard tradeoffs remain.
Audience takeaway: Concrete questions to settle on data access, cost controls, and ownership before scaling.
Shared AI Infrastructure Across Institutions
Build, buy, or share: how UC San Diego runs a multi-tenant AI platform supporting UC Berkeley, UC Merced, UC ANR, and Fairleigh Dickinson, and how shared services change the economics of campus AI.
Audience takeaway: How to weigh shared university services, local hosting, and commercial cloud against institutional budgets.