Crossing the Innovation Valley of Death: Democratizing Data and Artificial Intelligence for Research Administration
What happens when a Carnegie R1 institution and a primarily undergraduate institution recognize they're facing the same problems — and decide to build something together?
This keynote traces the origin story of a National Science Foundation-funded initiative that is redefining what's possible for research administration offices of any size or structure. The presenter will share how conversations between the University of Idaho and Southern Utah University revealed a common thread running through their institutions and many others: an increasing federal compliance burden, aging information systems, data silos, and a persistent inability to hire, build, or buy their way out.
From that shared frustration came a bold collaborative vision — to develop open-source, AI-powered tools, a unified data model that could be freely adopted, adapted, and scaled by the broader research administration community, and a community of practice to enhance research administration adoption and use of these and other tools.
Attendees will learn how the project team approached the challenge of introducing artificial intelligence into research administration responsibly. The session will introduce the TaMPER Framework (Task, Model, Prompt, Evaluation, Reporting) — a structured approach to deploying AI with transparency, reproducibility, and auditability. Real-world use cases will illustrate how discrete research administration tasks can be automated, connected into intelligent workflows, and evaluated.
The keynote will close with a forward-looking discussion of how both responsible use of AI and a unified data model — one that normalizes and connects data across institutions regardless of financial system, grants management platform, or institution type — could fundamentally transform research administration and democratize access to research intelligence, with an intention to level the playing field for under-resourced offices.
Attendees will leave with a clear picture of what this movement could mean for the future of the profession and how they can get involved through a growing community of practice.
By the end of this session, attendees will be able to:
1. Describe the landscape of systemic disparities in research administration infrastructure across institution types and explain why AI and a unified data model are being proposed as equitable, scalable solutions.
2. Apply the TaMPER Framework (Task, Model, Prompt, Evaluation, Reporting) to evaluate and responsibly deploy AI tools within their own research administration workflows.
3. Distinguish between high-impact and low-value AI use cases in research administration by identifying the characteristics of tasks that are most suitable for automation — including volume, risk level, and data sensitivity.
4. Articulate the value of a unified, open-source data model for research administration and explain how interoperability across institutional systems could transform data-driven decision making in the profession.
5. Identify concrete steps to engage with AI tools and a growing community of practice to bring AI-enabled research administration capabilities to their own institutions.