Finding your path forward with AI.
Stellic works with more than 200,000 higher ed staff across 100+ institutions. In addition, we surveyed registrars, advisors, enrollment leaders, and IT directors to understand how AI is landing in their day-to-day work. From inside those workflows, here's what we're seeing.
"We're starting with the most time-consuming, lowest-value parts of the admissions cycle. Transcript review, documentation, the work that doesn't require judgment. AI can handle that. Getting it right there feels like the right place to begin."
Enrollment Manager, Mid-Sized Private Institution
"AI gets sold as something that runs itself. What I know from implementing tools is that it needs maintenance, monitoring, and a clear owner. We're figuring out what that structure looks like before we say yes to anything else."
IT Director, State University System
"The university is saying use AI to save time, but not explaining how. So we're figuring it out for ourselves. Everyone's doing something a little different, and nobody's really decided what good looks like yet."
Advisor, Regional University
"We're using AI tools. The harder question is how to operate AI across the institution, not just use it. What workflows does it belong in, what guardrails do we need, what are we actually optimizing for. That's the part nobody's figured out yet."
Academic Leader, Research University
The above points to one problem: the work of AI adoption is already happening, and no one is helping them do it well.
A short assessment places you on the AI fluency curve and shows what to do from there.
Nearly 9 in 10 higher ed staff work at institutions already investigating AI.
Institutional posture has shifted faster than the public conversation suggests. More than half of respondents describe their institution as actively investigating AI tools, and another third say their institution is interested but proceeding cautiously.
The "should we?" question has largely been settled at the leadership level. What hasn't been settled is the "how should we?" question, and where active investigation leads varies widely by institution, role, and what people think AI is even for.
Higher ed staff use AI at work regularly.
Across registrars, advisors, IT, enrollment, and student success roles, AI tools have moved past the experiment phase. Nearly half of respondents use them daily, and the behavior is no longer concentrated among a few early adopters at a handful of progressive institutions. It's broadly distributed across roles, regions, and institution types.
Whether the tools being used are the right ones for the work, and whether the people using them have the framework they need, matters more than whether AI is being used at all.
The top worry isn't job loss. It's accuracy.
When we asked staff to name their concerns, the top answer wasn't the one that dominates conference panels. Accuracy of AI output led every other category by a wide margin, with student data privacy close behind.
Job displacement and over-automation are present in the responses, but they don't lead. The dominant concerns are operational: will the output be right, will student data be safe, will the institution be ready to use this responsibly. What staff are most worried about is AI being wrong about students at a scale that's hard to catch.
Reporting is where the appetite is sharpest.
The appetite is concrete. Staff aren't asking for general AI capabilities or a platform that does everything. They're naming specific bottlenecks — pulling data, surfacing patterns, reducing repetitive cycles — where small reductions in friction would meaningfully change how their week feels.
Reporting and data analysis led by a wide margin, which says something important about where staff feel most stretched and most ready for relief.
The field is split on what AI is for.
Underneath the operational findings is a more fundamental split. Just under half of respondents framed AI as something to handle routine tasks so they can focus on higher-value work. About a third framed it as a tool that enhances what they do.
Both are valid approaches, but they lead to different strategies, different tool choices, and different conversations about what AI is for on your campus. Most institutions haven't gotten that clarity yet, and that split is the bridge into the next section.
The biggest gap is training, not enthusiasm.
People are ready. What's missing is the shared framework — what good use looks like, what the guardrails should be, which workflows AI belongs in at all. Without that, people develop AI habits on their own, some well-considered and some quietly risky, and those habits become institutional defaults.
Want more training on how to use AI effectively at work.
The institutions that close this gap won't necessarily have better tools. They'll have people who know how to use whatever tools they have well.
The field is engaged, curious, and largely on its own. What's missing is a framework.
A short assessment places you on the AI fluency curve and shows what to do from there.
A way to think about AI that fits the work.
The default playbook is to do what you already do, just faster. Answer more emails. Process more requests. That's the 2x path, and it leads to exhaustion before it leads to progress. The 10x path starts with a harder question: which parts of your work require you at all? When you get clear on that, AI stops being a productivity tool and starts being something more useful — a way to reclaim your time for the work that only you can do.
The framework below is what we use at Stellic to think past it — for our own work, and with the institutions we partner with.
- Add AI on top of existing workflows.
- Optimize for output volume and speed.
- Linear gains that plateau quickly.
- Complexity and effort grow over time.
- Rethink which tasks require you at all.
- Narrow to the 20% that creates 80% of the impact.
- Exponential gains that compound over time.
- Simpler processes. More meaningful work.
- AI bolted onto the existing audit workflow.
- Decisions per hour go up.
- Same overrides, same edge cases, same exception rate.
- Less time per case means less time to read each one carefully.
- AI pattern-matches across articulation agreements.
- Policy edge cases get your attention.
- Equity considerations get real review time.
- Conversations with departments about why a rule exists.
- Routine outreach goes out faster.
- The calendar gets tighter.
- Caseload grows to fill the freed-up time.
- Less thinking room between meetings.
- AI handles routine nudges, reminders, and follow-up at scale.
- High-volume outreach goes out without eating advisor time.
- Your hour with the at-risk student is the work itself.
- Friction around the conversation falls away.
- AI speeds up document review and inquiry triage.
- Funnel feels lighter for a quarter.
- Yield numbers don't move.
- Board still wants more — same playbook, bigger pipeline.
- AI handles high-volume, low-context work.
- FAQ replies, document checks, routine outreach automated.
- Team time goes to counselor relationships.
- Community partnerships that shape the class profile.
- Every department has its own AI vendor.
- Procurement queue grows, integration debt piles up.
- Security review backlog stretches into quarters.
- Speed-of-approval becomes the metric.
- AI moves from procurement problem to governance problem.
- Institution publishes the criteria upfront.
- Vendor conversations run through that filter.
- Human-review boundaries are set before deals close.
The 10x question is harder: which parts of your work require you?
A short assessment places you on the AI fluency curve and shows what to do from there.
Start with the goal, not the tool.
AI is now a factor in technology decisions at nearly 9 in 10 institutions. These conversations are uncharted territory for most, so here's how we think about it at Stellic. Three principles to consider as you build your own:
The most durable AI investments make what you already have more powerful rather than asking you to start over. A tool that reasons over your actual student records, degree requirements, and institutional history will always outperform one generating outputs from generic training data. Start with the friction in your existing workflows. The technology should fit around that, not the other way around.
Some decisions in higher ed carry real weight for students — financial aid, academic standing, path to graduation. Those require human judgment and human accountability, and good AI should make that judgment better informed and faster to act on. Before adopting any tool, it's worth knowing: can staff see why a recommendation was made, and can they override it when they need to? If the answer to either is no, that's worth taking seriously.
The most common mistake in AI adoption isn't choosing the wrong tool. It's choosing a solution before the problem is well-defined. The most durable investments start with a specific bottleneck — something that slows down good work, creates friction for students, or costs more staff time than it should. When AI is the answer to a real question, it tends to stick. When it's added for its own sake, it tends not to.
Knowing what to look for in tools is half of it. Knowing where you stand today is the other half.
A short assessment places you on the AI fluency curve and shows what to do from there.