In 2008, Ultimate Medical Academy moved its healthcare training online at a time when almost no one was teaching healthcare that way. UMA is now the largest part of Vocate Education Solutions, a national nonprofit that trains more than 25,000 students a year, most of them working adults. Alexandra Schaffrath, previously Vocate’s chief financial and strategy officer, was recently named its president and CEO.

Schaffrath is the 14th guest on The New Normal, OutRival’s interview series with leaders deploying AI in production. The conversation covers how UMA built its organization around job placement, where automation helps students who study after midnight, and why she thinks institutions that plan year to year will not last.

Here are the five biggest takeaways from her conversation with OutRival co-founders Ruben Harris and Timur Meyster.

1. Outcomes decide where the staff goes

UMA’s accreditor sets three requirements for every cohort:

• At least 70% of students must be retained.
• At least 70% of graduates must find jobs in the field they trained for.
• At least 70% of graduates pursuing a credential must pass the exam.

UMA graduates 12,000 to 13,000 students a year, so the placement requirement alone means thousands of people have to land jobs in their field every year. Schaffrath says the career services office at the university she attended had about five people. UMA’s has 450, because, as she puts it, “our students come for a job.”

Almost 80% of UMA’s 3,000 employees work directly with students, from enrollment through job placement. Many of those students tried community college or another school first and didn’t succeed, through no fault of their own, because they never had that kind of support.

“A day in the life of some of our students would be a year for some of us,” Schaffrath says.

2. Data stands in for seeing the student

UMA’s students learn online, so staff can’t tell from a glance in the hallway whether someone is having a hard time.

“The data is the blood that runs through the organization,” Schaffrath says.

When a student starts missing classes or their grades begin to slide, those signals tell staff that something may be going on in the student’s life, and UMA uses them to decide when and how to step in.

Over the years, that work has given UMA a detailed understanding of its own processes and how each one affects students. Schaffrath says knowing its processes that well is what will let UMA replicate them in technology.

“If you don’t know a question, it’s hard to put an agent to work,” she says. “You don’t know if the answer is right.”

Subscribe to The New Normal here.

3. Support for students who study at 2am

UMA has staff available for most of the day but not around the clock, and many of its students do their coursework at 10pm or midnight, when they finally have the time.

“We don’t want them to feel like they’ve hit a wall at 2 in the morning,” Schaffrath says.

Her plan is to use agents and self-service tools to help students get past academic hurdles during those hours, such as a chapter they’re stuck on. Interview practice is another use. Students spend a lot of time preparing for job interviews, and instead of waiting for a staff member to be free, Schaffrath would like them to be able to run 50 practice interviews with an automated tool. UMA is building those tools now.

4. Employers want different things from administrative and clinical graduates

UMA works with thousands of employers, and what Schaffrath hears from them depends on whether they are hiring for administrative or clinical roles.

Employers filling administrative roles want to cut the cost of low-value work, which means some of the tasks UMA graduates handle today will be automated.

Schaffrath’s answer is to train students for where those jobs are heading. “It’s not about the task today. It’s preparing your students for the job of tomorrow,” she says. For that reason, UMA has added AI to its new curriculum, with less emphasis on any single tool and more on how students approach AI, use it, and check whether its output is right.

Clinical employers, by contrast, expect AI to add to the demand for their workers.

Patients in a health crisis still want a person caring for them, and Schaffrath hears that AI is creating more opportunity for that care. On a tour of Tampa General Hospital’s cardiac catheterization lab, where doctors use imaging to diagnose and treat heart conditions, she saw AI put scans together fast enough that technicians no longer have to read every one, which lets the lab see more patients. She expects UMA’s clinical graduates to get busier as hospitals use AI to provide more care.

Across both groups, the request she hears most often is for flexible workers, since employers can’t predict what a given job will look like in two or five years.

5. Annual planning cycles are over

Schaffrath believes slow decision-making puts an institution’s future at risk. “We cannot live in cycles that are annual anymore,” she says, and in her view, the institutions that wait to see fall enrollment before making decisions will not survive.

UMA has had to reverse some of its own decisions. In 2008, no off-the-shelf product fit its model, so it customized its core systems. That customization has since become expensive to maintain, and with commercial technology catching up, UMA now plans to move to standard systems that it configures to its needs.

Inside the organization, Schaffrath lets staff run pilots as long as they state what they expect to learn and set a time limit. Placement results take 18 months to show up, so teams running those pilots look for earlier signs that something is working. “As long as we agree on the guardrails, I don’t think we need to throttle the innovation,” she says.