Earlier this spring, the emergency department at the Cleveland Clinic told Dr. Rita Pappas it had a problem. 

The Cleveland Clinic is the largest hospital system in Ohio, but after an influx of patients came in, the emergency department did not believe it could accommodate all the people who would need care. 

Pappas, the Cleveland Clinic’s medical director for the admission and transfer center, disagreed. 

The Cleveland Clinic was in the midst of trying new AI-powered software from Palantir. The system was predicting that there would be a large number of discharges that day, meaning there would be enough space for all the incoming patients. 

Pappas and her team convinced the emergency department to give Palantir’s platform a chance, and — to her relief — the new system worked. 

“It was perfect. We were able to accommodate all the patients, and so everyone was cheering, they were so happy,” Pappas, who also practices clinically as a pediatric hospitalist, told CNBC in an interview. “Everyone got really excited. I was very excited because I do not want to have the emergency department yelling at me.”

Ordinarily, Pappas said managing patient flow in a similar scenario would require constant communication, every hour throughout the night. But by accurately predicting patient discharges, Palantir’s system saved Pappas, her team and the emergency department a lot of time and effort.

Palantir partnered with two health-care systems, Cleveland Clinic and Tampa General Hospital in Florida, during the summer of 2021 to develop software called Palantir for Hospital Operations. Now, just two years later, the company says its hospital-operations platform accounts for around 10% of its commercial revenue in the U.S.

The software translates complex data into digestible information that health-care workers can use to guide their decision-making and resource allocation. It can be used to generate real-time and predictive insights into areas like staffing, wait times and hospital-bed assignments.

Shyam Sankar, the company’s chief technology officer, said he thinks Palantir for Hospital Operations is just getting started. 

“I’m so excited that we’re at 13% of beds in the U.S., but you know, like 90% of the work is in front of us here,” Sankar told CNBC in an interview. 

Palantir specializes in data analytics, and shares of the company are up around 150% so far this year as bullish AI investors are buying into its tech offerings. The company reported its first quarterly net profit for Q4 2022, and Palantir CEO Alex Karp said in a May letter to shareholders that the company expects to remain profitable “each quarter through the end of the year,” and that demand for its new Artificial Intelligence Platform (AIP) is “without precedent.”

The company is perhaps best known for its work with the U.S. government’s defense and intelligence agencies, but despite its high-profile customers in government, Sankar said Palantir works to tackle hard challenges across a variety of industries, including health care — a field that experts widely believe is in crisis. 

Nurses and physicians are experiencing high levels of burnout, and staffing shortages across the country have made it difficult for hospital systems to keep up with patient demand. These challenges were magnified by the Covid pandemic, as nearly 63% of physicians reported symptoms of burnout in 2021, up from 38% the prior year, according to a recent study co-authored by the American Medical Association.

Companies across the medical and technology fields have been working to build solutions that could bring health-care workers some relief. But unless the new platforms, tools and services can be easily integrated into workflows, physicians and nurses often find they can be more trouble than they’re worth. 

“If you don’t build it in the workflow of the user, it actually doesn’t get used, it just sits on the shelf,” Dr. Peggy Duggan, executive vice president and chief medical officer of Tampa General, told CNBC in an interview.  

As such, in the summer of 2021, a team of four Palantir employees set out to create a meaningful solution that health-care workers at Cleveland Clinic and Tampa General could actually use. It was a daunting challenge, and there was an additional catch: They were only given a matter of weeks to build something worthwhile.

Drew Goldstein and Jeremy David led Palantir’s team of four, and they now run its fast-growing U.S. commercial health-care team. The pair said they were given around 12 weeks to build a solution for Tampa General, and their work in Florida was already underway when the Cleveland Clinic’s chief information officer called to give them an even shorter timeline: eight weeks. 

“For better or for worse, I’m quite hopeful and optimistic,” David said in an interview with CNBC. “I was like, ‘Eight weeks? We can get something live in three.'” 

David, Goldstein and their two teammates took an intense, hands-on approach to building their hospital-operations software. 

They spent their allotted time on the ground in the hospitals in both Ohio and Florida, shadowing doctors and nurses during their shifts, even overnight. They asked questions about existing workflows and tried to grasp the biggest pain points for each organization, an experience Goldstein said was both “overwhelming” and “exciting.” 

To their surprise, the Palantir team learned that hospital staff were carrying out some of their operations by manually entering data into spreadsheets and configuring information like scheduling on paper and whiteboards. 

The Palantir team also became acquainted with whiteboards, as Goldstein said much of their early efforts in Ohio were spent trying to draw out existing workflows with hospital-operations leadership. 

“It was so foundational in our understanding of what these problems could look like,” Goldstein told CNBC in an interview.

At the Cleveland Clinic, the Palantir team initially focused on addressing challenges related to patient flow, particularly around accelerating aspects of the patient-discharge process, while they trained their focus on clinical-staffing allocation at Tampa General. But before long, Goldstein said it became clear the health systems shared both of those problems.  

“So then, instead of just building these completely separately, we started to think about, like, what should the product look like to be able to solve this?” Goldstein said.

When their work was done in Florida and Ohio, Palantir’s team had what amounted to an early iteration of Palantir for Hospital Operations. David said it was the product of several 100-hour weeks where he and Goldstein, who have become good friends, often lived in adjoining hotel rooms. 

“Drew didn’t like that,” David joked. “[I’d] knock on his door, ‘Good morning, Sunshine!'” 

David said the team’s hands-on approach was essential in order to understand what support health-care workers truly needed. 

“If you treat the nurses and patients like widgets, you’re going to lose,” David said. “For about two weeks, we tried to treat them like widgets, and we lost.”

The group presented their work to hospital leadership through a series of demos and testimonials, and David said it was clear the health systems believed Palantir had created something special. 

Their next challenge, Goldstein said, was to keep building and execute on a broader, longer-term vision.