How Syracuse University deployed Claude to every student, faculty member, and staff

Changing how a university teaches and operates

Jeff Rubin, Senior Vice President and Chief Digital Officer, has taught at Syracuse for 30 years. The classroom model he wants to change has been in place, largely unchanged, for over a century: a professor lectures, students take notes, and assessments measure what students have learned. "Whether there are 200 students or 15 students, the problem is always reaching every student, where they are," Rubin said. "Some students believe the content is moving too fast, some too slow. You can't individualize it."

The rollout took two forms, each with its own challenge: on the adoption side, convincing faculty that AI belonged in the classroom. On the data side, preparing decades of institutional information so Claude could actually work with it.

In October 2025, Syracuse gave every student, faculty member, and staff member a Claude license and focused on training, governance, and the integration of AI into the classroom. At the same time, the data and AI team began building custom applications using Claude's models, connecting them to institutional data to support advising, fundraising, and operations. "AI is now at the center of our digital transformation," Rubin said. "I'm not sure there's anything we plan to do where AI doesn't play a role."

Selecting Claude for ethics and safety 

Syracuse evaluated multiple AI providers for more than a year before choosing Claude, and model quality was only part of the decision. "The ethical side, the safety, the values, those align with what we want to do within higher ed," Rubin said. "From the beginning, Anthropic looked at us as a partner, not a standard vendor relationship."

Andrew Joncas, the university's Assistant Vice President of Enterprise Data and AI, cited Anthropic's open publication of the MCP standard and its willingness to share unflattering test results as evidence of that alignment. When the team raised a need for enterprise-scale Claude Code quotas, Anthropic responded within a week with a beta program. "It's one thing to be heard and written down on a piece of paper," Joncas said, "and another where someone comes back and actually gives you a solution."

Making the university-wide rollout work

Once every student and faculty member had access to a Claude license, the question on the adoption side became how to make sure it was used well. Rubin started in his own classroom. He built a practice exam with multiple-choice questions, but students who used it scored only marginally above average. When he consulted experts at the School of Education, they told him multiple choice wasn't testing recall. Students were clicking through answers without engaging with the material.

So he redesigned the assessment. Similar to a flash card, students now receive a term and type in an answer. Claude grades their response and identifies where they could be stronger, based on actual words Rubin has used in class. Exam scores jumped 12 points, the highest first-exam results in his 15+ years of teaching the course. "It wasn't the tool," Rubin said. "It was how I was using it."

To bring the broader faculty along, the university launched voluntary AI workshops that now regularly draw 200 to 500 participants per session. They also created Communities of Practice: open forums where faculty and staff discuss concerns ranging from academic integrity to mental health. One faculty member who initially described AI as "ruining education" recently sent a note saying she was excited about the future. "I think it's because she was heard," Rubin said. "People are so focused on the students, they're forgetting: we've got to educate the educators."

Connecting Claude to institutional data

The second challenge was technical. The university's digital transformation touches everything from donor relations to class search to campus security. To support that, Syracuse’s data and AI team built an agentic platform for higher education, where the user might be a faculty member, a staff member, or a student on the same day. The platform runs on Opus 4.6, which the team found reliable across complex workflows requiring more than ten sequential steps. Claude orchestrates specialized agents that run Data Analysis Expressions (DAX) queries against Microsoft Fabric or retrieve data through custom MCP connections."

The most visible product of this work is Clementine, a natural language class search tool that launched this semester. The old system was keyword-based: a student typing "AI" would get painting classes because the letters appear in the word. Clementine uses Claude Opus 4.6 to query millions of rows of enterprise data, so a student can ask for an AI class on Wednesday afternoons and get results contextualized to their schedule and career goals. Later phases will integrate transcripts and graduation requirements. "Your data is what makes AI yours," Joncas said. "It differentiates your organization's abilities to do things that other organizations can't." 

Building the platform was the straightforward part. Getting the university's data ready for it was not. Few people fully understand the data relationships and rules that have accumulated over decades. "The groups we're working with have never been asked to define the university through its data and through its processes before,” Joncas said. His team addressed this by recording departmental meetings where staff debated the rules governing their own data and feeding the transcripts into Claude so it could learn the governing logic. 

One soon-to-be-released project gives deans access to donor reports filtered by geography and giving history in just minutes, and it took about three months of data preparation. Another project connects Claude to the university's security infrastructure, which ingests roughly a terabyte of log data daily. “We're just taking all the data, putting AI in front of it, and getting results in minutes,” Joncas said. 

Growing adoption across the institution 

On the adoption side, every category of user showed steady growth from October through April, with student peak daily active users up 394%, staff up 214%, and faculty up 146%. Adoption reached the top of the organization as well: 123 of 139 university leadership team members are active users. Ryan Elstad, Senior Director and Architect for the Data and AI team, pointed to the rapid adoption of Claude connectors like Microsoft 365, Confluence, and Jira. The majority of conversations now include file uploads.

On the technical side, the tools are spreading faster than expected. Claude Code saw immediate uptake from both developers and non-developers upon release. One classroom development specialist with no Python experience used it to prototype control system software. "I went by the conference room, and the monitor outside said 'Claude Code session on writing Extron control system software,'" Joncas recalled. "It had been one week."

A budget planning officer told Elstad that work that had previously taken hours now took minutes with Claude. "I asked, 'So did you go home early?'" Elstad said. "His answer was, 'No, I got more work done.'" Another staff member commented, "I can now make my vision happen,” describing their ability to convey an idea directly rather than waiting for an IT team to implement it.

Looking ahead, the class search and registration project has five more phases planned, and each major project is making the next one faster as departments across the university articulate their data rules for the first time. 

For Rubin, the momentum reinforces the original bet: that preparing students to work with AI is as important as any single product the university builds on it. "AI is going to be a life skill," Rubin said of the university’s decision to push forward with Claude. "When a professor redesigns an exam and scores jump 12 points, when a budget officer finishes in minutes what used to take hours, when a dean pulls a donor report that used to take two weeks; that's when you know it's working. AI isn't just a tool we hand our community, it's a skill we teach them to use well. That's what we're building toward across the university."