HCA Nurses Say Palantir’s Timpani Scheduling Tool Misses the Mark

HCA Healthcare has rolled out Timpani, an AI-assisted nurse scheduling system it co-developed with Palantir, at roughly 130 of its 190 locations since 2023. Nurses interviewed by WIRED say the system often assigns shifts that conflict with their requests and leaves them spending more time trying to appeal or trade assignments.

The core tension is practical: HCA says nursing leaders make the final decisions, while nurses describe a process in which changing a software-generated schedule can be difficult and late. That gap matters in a hospital, where staff say the mix of experience on a shift affects how care teams operate.

A schedule that starts with forecasts

Timpani runs on Palantir Foundry and uses algorithmic forecasts of patient volumes and staffing needs to generate nursing schedules. HCA launched it in 2023 with the stated aim of saving managers time and money and reducing favoritism in shift assignments.

One Florida critical care nurse, Amber Retzloff, told WIRED that over four months she requested 50 specific 12-hour shifts and was assigned to different shifts more than half the time. She said the resulting runs of consecutive workdays left her mentally drained. Other nurses described schedules that ignored preferences for spacing shifts, working nights or weekends, or taking particular days off. These are workers’ accounts, not a measured system-wide assessment.

HCA cites different measures of performance. The company says Timpani schedules nurses for an average of 1 percent of their requested days off. It also says the system has reduced managers’ scheduling hours and reliance on contract nurses, and increased retention. Those company-reported benefits do not answer whether individual assignments feel workable to nurses.

Human approval is not the same as easy correction

HCA spokesperson Harlow Sumerford says nursing leaders, not Timpani, make final scheduling decisions, and says the company is improving the tool based on nurse feedback. Nurses told WIRED that appeals now go to a centralized team at HCA headquarters, while shift swaps can be difficult; two nurses put their chance of success at about 50 percent.

The distinction is important for evaluating an AI-assisted workflow. A human may formally retain authority, but the schedule still shapes the starting point and the effort required to correct it. Nurses also alleged that schedules sometimes left teams short of experienced staff, raising concerns about patient care. The reporting does not establish that Timpani caused patient harm.

A former HCA data science manager made a separate, unresolved allegation in a lawsuit: that HCA routinely deleted data used to create schedules, making it harder to audit the system. At the time of WIRED’s reporting, HCA had not formally responded in court. If the underlying data cannot be retained and reviewed, it becomes harder to distinguish a forecast problem from a staffing-policy decision or a local override.

A separate Foundry tool, a related oversight question

WIRED also reported complaints at Rayus Radiology about a different Palantir Foundry-based tool, designed to populate scheduling fields from doctors’ orders. Workers said it sometimes linked requests to the wrong patient profiles, mismatched tests or scans, or generated personal details. Rayus disputed the broad characterization and said its technology supports, rather than replaces, appropriate human judgment and oversight.

The two systems perform different jobs, so their reported issues should not be conflated. But both accounts make the same operational point: software can present a recommendation or fill in a workflow, while the quality of human review and the ability to correct errors determine what happens next. At HCA, the unresolved test is whether nurses can meaningfully challenge schedules before the burden of fixing them lands on the floor.