Nurses working at HCA Healthcare are challenging an AI-assisted scheduling system that the hospital operator presents as a way to match staffing with patient needs. In an October 2 report, WIRED described complaints about uneven experience levels, unwanted shifts and difficulty securing corrections. HCA disputes the characterization that its technology sacrifices care and says nursing leaders retain final authority. The disagreement concerns more than whether software can assemble a rota: it concerns what happens when staff believe a technically generated schedule does not reflect the practical demands of a hospital floor.
Nurses describe a gap between schedules and working conditions
HCA’s Timpani system, developed with Palantir, has been introduced at roughly 130 of the company’s 190 locations since 2023. It uses patient-volume forecasts and staffing information to produce schedules. Nurses interviewed by WIRED say the resulting assignments can leave shifts short of staff or with too few experienced colleagues, particularly on Sundays. They also describe preferences about nights, weekends and time off being disregarded.
One nurse and union leader, Retzloff, recounted working alongside four junior colleagues and delaying attention to the sickest patients while helping less experienced staff. That account illustrates the nurses’ concern about the distribution of experience, not simply the number of people assigned. WIRED spoke with five other nurses who raised similar scheduling objections. Their testimony describes perceived risks and working conditions; it does not establish an independently measured causal link between Timpani and patient harm.
HCA spokesperson Harlow Sumerford told WIRED that nursing leaders make the final decisions. The company denies designing the system to cut staffing costs at the expense of care and says its purpose includes aligning skills, patient needs and employee preferences. It also says the system is being improved through feedback. Those assurances leave the central disagreement unresolved: staff describe constraints they cannot readily correct, while HCA describes a tool operating under human supervision.
The route for corrections is part of the dispute
According to nurses in the report, local managers initially adjusted schedules but appeals now go to a centralized HCA team in Nashville. They say this makes it harder to explain individual circumstances directly. When a request is rejected, arranging a swap with a colleague can become the remaining option. The reported difficulty is therefore partly about the process surrounding the model, rather than only the model’s initial output.
HCA says nurses are scheduled on just 1 percent of days they designate as unavailable. Missouri nurse and union leader Lee Barker described the change differently, saying assignments on requested days off had previously been unusual and now affected many colleagues. The company-wide figure and the personal accounts address different levels of experience. Neither, on its own, answers how inconvenient assignments are distributed or how successfully exceptions are handled.
The company has also presented favorable performance results. A July paper cited reduced scheduling work for managers, lower reliance on expensive contract nurses, improved retention and a 98 percent rate for achieving the intended mix of skills and experience. These are HCA’s reported findings, not independent verification of the nurses’ complaints. National Nurses United, which represents around 10,000 of HCA’s 100,000 nurses, says requests for information about the system have been rejected.
Oversight depends on records as well as final approval
A separate dispute concerns the ability to examine how Timpani operates. Former HCA data-science manager Angelique Russell alleged in a July lawsuit that she was dismissed in retaliation for raising concerns about the system. Her allegations included routine data deletion that could impede auditing and potentially breach healthcare requirements. HCA had not formally answered the lawsuit when WIRED reported it, and Russell declined to comment. The allegations should not be treated as established violations.
WIRED also described problems involving a different Palantir-based tool at Rayus Radiology, including erroneous patient and test information in internal records. Rayus disputed the broader characterization and said the technology supports people rather than replacing them. That separate deployment does not demonstrate that Timpani has the same defects.
For HCA’s scheduling dispute, the immediate issue remains the distance between formal human authority and a nurse’s ability to obtain a workable change. Human approval is one control; accessible correction procedures and records that permit scrutiny are others. The report supplies competing accounts of those controls, not a settled verdict on the system’s safety.
