Case Study
Ethical AI on Union Theological Seminary’s Terms
How a mission-driven institution and a family-owned AI company are exploring technology deliberately — full data ownership, agents that read kept separate from agents that write, and a human approving anything sensitive.
Union needed to reach prospective and current students more effectively — but responsibly, affordably, and with leadership oversight intact. So the partnership starts with an assessment phase rather than a deployment, connecting to systems the seminary already runs instead of replacing them.
Read ≠ write
Agent roles deliberately separated
Human approval
Required at sensitive touchpoints
Existing systems
Connected, not replaced
Family-owned
Partner based in New York, NY
Bridging mission and innovation
Union Theological Seminary, one of the most prominent theological institutions in the United States, faced a question many mission-driven institutions share: how do you explore emerging technology while staying true to your values? For an institution formed around preparing ethical leaders, AI could not be approached casually.
Any technology partnership had to prioritize four things from the outset:
Student privacy and data sovereignty
Where the data lives, and who can reach it, is a values question before it is a technical one.
Transparency in how information is used
Staff and students should be able to see what the system does with what it knows.
Human oversight at every step
Automation that acts on a student's behalf needs a person accountable for the action.
Efficiency that serves people
The goal is freeing staff for student support — not replacing the humans who provide it.
A partner, not a vendor contract
When Donald Joshua, Union’s CIO, began exploring AI possibilities, the fit became clear. ibl.ai is family-owned and operated from New York, NY — not a faceless conglomerate. Leadership is directly accessible, and decisions are made deliberately rather than driven by outside investors.
Full code and data ownership
Union keeps complete control over its data and infrastructure. No black boxes, and no data leaving without explicit permission.
Human-in-the-loop architecture
Agents that read data are deliberately separated from agents that write data, which forces a human review at the decision points that matter.
Transparency by design
Every workflow is visible, auditable, and aligned with the institution's own governance rather than a vendor's defaults.
Three phases, in order, on purpose
Rather than rushing to deploy, the partnership begins with a deliberate exploratory phase — connecting to what exists before deciding what to build.
Assessment and access
Connect with the systems already in place — Formstack, Sonis SIS, Moodle LMS — rather than replacing them. Understand the API capabilities and data flows that exist, and determine what is possible while maintaining security and privacy.
Secure infrastructure
Build Model Context Protocol layers that give agents controlled access to institutional data through those existing platforms, establish clear boundaries between read and write operations, and create audit trails for every AI interaction.
Collaborative workflow design
Work with academic leadership and staff to define how agents should behave, ensure every automated communication reflects Union's voice and values, and keep human approval at sensitive touchpoints so leadership stays in control.
Donald Joshua
Chief Information Officer, Union Theological Seminary
“We’re trying to make ourselves more efficient... but we’re looking at it very ethically. We’re trying to be very diligent about how we do it and ultimately respectful of everybody’s privacy.”
This is not AI for AI’s sake. It is a careful, values-aligned exploration of what becomes possible when a family-owned technology partner works alongside a mission-driven institution.
Engaging students responsibly
Timely, relevant communication with prospective and current students — without intrusive data practices.
Marketing affordably
Less reliance on expensive external platforms, while keeping the personal touch that defines the institution.
Full leadership oversight
Every automated process stays visible and controllable by Union's own team.
Less administrative burden
Staff time moves back toward student support, which is the work the seminary exists to do.
Explore AI without giving up control of it
Connect the systems you already run, keep the read and write paths separate, and put a human in front of anything that touches a student — on infrastructure you own.