Why automation keeps failing , and it’s almost never the robot’s fault

In early 2025, MultiCare Health System in Tacoma, Washington pulled the plug on fourteen Moxi robots it had deployed across its hospitals. The five-foot, 300-pound logistics bots , equipped with blinking blue eyes that turned heart-shaped when they encountered staff , had been rolled out just two years earlier with considerable fanfare. Nurses complained they got in the way. Administrators said the economics didn’t stack up. The program was shuttered.

Coverage was swift and predictable. For skeptics, it was confirmation that robots don’t belong in hospitals. For boosters, it was a speed bump on the road to automation. Both camps largely missed the real story: Moxi, by every technical measure, worked. It completed over a million deliveries across its fleet, saved pharmacy teams thousands of hours, and earned genuine affection at the sites where it thrived. The problem at MultiCare wasn’t the robot. The problem was that MultiCare , like so many organizations rushing to automate , never asked whether it was ready for a robot.

This is the central delusion of modern automation strategy: the belief that robots are essentially plug-and-play. That you buy the hardware, connect to Wi-Fi, and watch inefficiency evaporate. It is a myth with a very high price tag.

“The robot may be ready. The crucial task is ensuring the environment is too.”

The Appliance Fallacy

We have been conditioned to expect technology to conform to us. We plug in a toaster; it toasts. We install software; it runs. This habit of thought , call it the Appliance Fallacy , has migrated into how organizations think about robotics. Buy the machine, deploy the machine, harvest the efficiency gains. The reality of a robot operating in a live human environment is fundamentally different.

A hospital is not a laboratory. It is a socio-technical system built over decades around human intuition, improvisation, and organic, often undocumented workflows. Hallways get crowded. Shift changes create chaos. Code blues happen at the worst possible moment. Elevators are old. Inventory systems exist only in people’s heads. When a robot enters this environment, it doesn’t just automate a task , it collides with every hidden assumption and workaround that the organization has accumulated over years.

The contrast with industrial robotics is instructive. Robots have transformed automotive manufacturing precisely because the environment was redesigned around them first. Factory floors are structured, standardized, and largely free of unpredictable human movement. There are no patients having a crisis in the middle of the assembly line. The robot’s ‘operating system’ and the facility’s ‘operating system’ are aligned. In healthcare, that alignment is almost never a given , and rarely even a goal of the procurement process.

When the Environment Breaks the Robot

To understand how environment defeats technology, consider a deceptively simple task: a logistics robot delivering stat medications from a basement pharmacy to the ICU during the 7 AM shift change. The robot is technically capable. Its navigation algorithms are sound, its sensors calibrated, its software up to date. Then reality intervenes.

The elevator is analog , no digital interface means the robot waits indefinitely for a human to press the button. The hallways are packed; the robot’s safety protocols cause it to stop repeatedly, turning a two-minute delivery into a fifteen-minute crawl. When it finally arrives at the ICU nurses’ station, no one is there , a Code Blue is underway , and the medications sit locked in the robot’s compartments, chirping politely, because there is no smart locker for autonomous drop-off. Three separate failure modes, none of them the robot’s fault, all of them entirely predictable to anyone who had studied the environment in advance.

Research consistently confirms this pattern. A 2024 literature review in the journal Healthcare found that the barriers to hospital robotics deployment cluster around infrastructure gaps, workflow misalignment, and staff cultural resistance , not robot capability. A qualitative study in the Journal of Medical Internet Research found that robots designated for controlled environments like pharmacy automation faced ‘fewer challenging sociotechnical implications’ than those operating in open human spaces. The more human contact, the more complex the failure modes.

The MultiCare case reflects exactly this dynamic. Where Moxi thrived , at Shannon Health, Mary Washington Healthcare, UTMB Angleton Danbury , it was integrated into structured workflows, supported by dedicated operational ownership, and embraced by staff who understood its capabilities. Where it failed, the organizational ground had not been prepared. The same robot, radically different outcomes.

The Five Dimensions of Readiness

The technology industry has long used Technology Readiness Levels (TRLs) to assess how mature a given system is for deployment. What it has largely lacked is a corresponding framework for the receiving environment , a way to ask not ‘is the robot ready?’ but ‘is this organization ready for a robot?’ Five dimensions define this readiness, and gaps in any one of them can undermine even the most technically sophisticated deployment.

1. Physical and Digital Infrastructure

Hallway widths, elevator interfaces, door systems, badge access, Wi-Fi coverage, charging station placement , these details seem mundane until a half-inch floor transition causes a robot to stall seventeen times per shift. Infrastructure readiness means auditing the physical and digital fabric of a facility before procurement, not after. Foster + Partners, in their work designing next-generation hospitals including the new Mayo Clinic in Rochester, Minnesota, now argue that architectural decisions must account for robotic integration from the very first drawing , because retrofitting infrastructure is exponentially more expensive than designing it in.

2. Workflow Standardization

Humans are remarkable at improvising. Robots cannot improvise at all , and this is actually a gift, because it forces organizations to confront how much of their operation runs on informal workarounds. Successful robot deployments almost universally require redesigning the process first, then automating the redesigned process. The robot cannot find a missing delivery bin if the inventory system exists only in a senior technician’s memory. Workflow readiness means documenting, standardizing, and digitizing the handoffs before the machine arrives.

3. Human Integration and Cultural Acceptance

No software patch fixes a nurse who views a robot as an obstacle to patient care rather than a collaborator. Cultural readiness is perhaps the dimension most consistently underestimated in procurement conversations. Staff need to understand what the robot does, what it doesn’t do, who owns the process when something goes wrong, and , critically , that it is not coming for their jobs. The healthcare systems where Moxi succeeded invested heavily in change management, training, and authentic communication. Those that treated it as an IT deployment rather than an organizational transformation largely did not.

4. Operational Ownership and Exception Management

Every robot will encounter exceptions. A blocked pathway. An unexpected input. A hardware glitch at 3 AM. Operational readiness demands clear, pre-established answers to fundamental questions: Who is responsible when the robot stops? Is it IT, facilities, operations, or the nearest nurse? How quickly can issues be escalated? Who has the authority to override? Without this clarity, robots fall into organizational gaps , technically functional but operationally stranded, accumulating resentment instead of ROI.

5. Economic Realism

Vendors and internal champions alike tend to present automation ROI in the most optimistic possible light. Year-one projections routinely ignore workflow redesign costs, infrastructure upgrades, change management investment, ongoing technical support, and the inevitable learning curve. Economic readiness means building financial models around realistic timelines and acknowledging that early deployments may be primarily learning investments rather than efficiency harvests. When organizations expect 300% returns in twelve months and receive a complex operational transition instead, even a technically successful robot gets discontinued.

The Coming Reckoning

The global surgical robotics market was valued at $11 billion in 2024 and is projected to approach $29 billion by 2032. The International Federation of Robotics reports that medical robot sales grew 36% in 2023 alone. Capital is flooding into the space, vendor promises are growing bolder, and health systems , facing endemic staffing shortages, rising costs, and relentless performance pressure , are increasingly receptive. The conditions for a wave of high-profile failures are forming.

The irony is that this wave is largely avoidable. The Moxi story is ultimately not a cautionary tale about robots , Diligent Robotics was acquired by Serve Robotics in January 2026 and is actively deploying its next-generation Moxi 2.0 platform to an expanding hospital network. It is a cautionary tale about organizational readiness. The same technology that stumbled in Tacoma is saving hundreds of thousands of hours of nursing time at facilities that did the work of preparation.

The ask is not that organizations stop pursuing automation. The labor math in healthcare alone makes continued investment in robotics near-inevitable: nurses spend up to 30% of their shift on non-clinical tasks; the nursing shortage is structural and deepening; robotic assistance in logistics, pharmacy, and materials management can meaningfully address both. The ask, rather, is that organizations stop treating ‘deploy a robot’ as the end of the conversation and start treating it as the beginning.

That means conducting honest environment audits before procurement. It means involving frontline staff , the people who will actually work alongside the machine , in deployment design rather than informing them after the fact. It means establishing genuine operational ownership structures, not vague cross-departmental responsibility. It means building financial models that account for the full cost of integration, not just the sticker price of the hardware. And it means accepting that the first deployment in an organization is almost always a learning investment, full stop.

The Smarter Question

For decades, the dominant question in automation has been: ‘Is the robot ready?’ It is the wrong question, or at least an incomplete one. The right question , the one that separates organizations that successfully harness automation from those that generate cautionary case studies , is: ‘Are we ready for the robot?’

Answering it honestly requires a kind of institutional self-examination that procurement cycles rarely accommodate and organizational culture rarely rewards. It requires admitting that workflows are messier than they look on paper, that staff relationships with new technology are complicated and consequential, that a half-inch floor transition matters, and that the real work of automation is organizational change, not machine installation.

The robots are getting smarter, faster, more adaptable, and cheaper every year. The environments we deploy them into are changing far more slowly. Until organizations close that gap , until readiness assessment becomes as standard a part of automation strategy as vendor evaluation , the myth of the plug-and-play robot will continue extracting its toll in wasted capital, squandered potential, and the enduring suspicion that the machines just don’t work.

They do work. We just have to be ready to let them.

Research sources: MDPI Healthcare (Dec 2024), Journal of Medical Internet Research, Diligent Robotics / Serve Robotics (2025–2026), DelveInsight Surgical Robotics Market Report, International Federation of Robotics, California Association of Healthcare Leaders, Foster + Partners Hospital Design Research, Nurse.org investigative reporting on MultiCare Health System.