AI & Infrastructure
Aug 2026
·
6 min read
Zohar Sherman
CTO deliverz.ai
Artificial intelligence in healthcare is often associated with clinical applications -diagnostics, imaging, decision support, documentation, drug discovery, and increasingly generative AI. But hospitals don't operate on information alone. Every day, thousands of physical tasks must also happen: medications need to reach patients, specimens need to reach laboratories, equipment and supplies need to move between departments, patients need transportation, and staff and resources need to be coordinated across increasingly complex facilities.
This is where Physical AI represents the next evolution of healthcare automation: applying intelligence not only to information, but to how work gets executed in the physical world.
Generative AI primarily creates and processes information; Physical AI connects intelligence with action in the physical world.
Hospitals generate thousands of physical operational decisions involving medications, specimens, supplies, equipment, patients, staff, and infrastructure.
Autonomous robots are one component of Physical AI, but the larger opportunity is coordinating robots, people, equipment, infrastructure, and workflows.
AI can make hospital operations more dynamic and predictive by anticipating demand, identifying bottlenecks, and reallocating resources.
The goal isn’t simply faster deliveries - it’s creating more predictable, efficient, and resilient hospital operations.
Physical AI applies artificial intelligence to systems that perceive, decide, and act in the physical world.
While generative AI helps people create, interpret, and process information, Physical AI connects intelligence with physical execution.
In healthcare, this can include autonomous mobile robots, robotic systems, intelligent infrastructure, sensors, and the software that coordinates them with hospital employees and workflows.
The distinction is important. A hospital doesn’t simply need to know that a medication is ready. It needs to determine when it should move, how urgently it needs to arrive, which resource should transport it, what route that resource should take, and how that task should be prioritized against everything else happening in the hospital.
Healthcare IT News described this as an emerging “hospital operations intelligence” layer - technology capable of continuously coordinating resources according to factors such as urgency, location, workload, predicted demand, and operational priorities. healthcareitnews.com
Healthcare AI has understandably concentrated on clinical information and decision-making. But hospitals are also enormous physical operations.
Every day they make thousands of decisions involving moving chemotherapy medications, laboratory specimens and equipment; assigning patient transportation; dispatching staff; coordinating elevators; and balancing workloads between departments. Many of these decisions remain manual, reactive, and disconnected. healthcareitnews.com
That creates an opportunity for AI that goes beyond assisting clinicians with information.
Physical AI can help answer operational questions continuously:
What needs to happen next?
How urgent is it?
Which resource is available?
Where is that resource?
What else is competing for it?
What decision will produce the best operational outcome?
That is a fundamentally different role for AI.
The robot is a powerful resource, but Physical AI is ultimately about deciding how every available resource should be used.
Hospitals may already have autonomous delivery robots, automated pharmacies, pneumatic tube systems, human transport teams, elevators, access-control systems, and numerous digital platforms.
Each may perform its own function extremely well.
The problem is that they often operate independently. Healthcare IT News describes this fragmentation as the missing link in hospital automation: hospitals don’t necessarily need to replace existing investments - they need to make those investments work together. healthcareitnews.com
For example, an urgent specimen may be best transported by a robot. A patient transport request may require a trained human transporter. Another delivery may be assigned to whichever resource can meet the required SLA most efficiently.
Physical AI therefore shifts the question from:
“What can this robot do?”
to:
“What needs to be done, and what is the best available resource to do it?”
Physical AI can move hospital logistics from reactive dispatch toward predictive operations.
Traditional logistics responds to events as they happen: a request arrives, someone assigns it, and a person or robot performs the task.
AI creates the possibility of learning the hospital’s “operational pulse.” By analyzing recurring demand, congestion, resource utilization, departmental behaviors, and other operational patterns, systems can begin anticipating what is likely to happen next. healthcareitnews.com
A system could identify an approaching demand peak, anticipate a bottleneck, reposition available resources, or rebalance workloads before service levels deteriorate.
That changes automation from simply executing tasks to continuously optimizing operations.
The objective of Physical AI isn’t to remove humans from healthcare.
It is to use people more effectively.
Success should therefore not be measured simply by how many robots a hospital deploys or how many deliveries they complete.
More meaningful questions are:
Can nurses spend more time with patients rather than transporting medications or specimens?
Can patients begin treatment sooner?
Can laboratories receive specimens more predictably?
Can operating rooms experience fewer logistics-related delays?
Can hospitals handle increasing demand without proportionally increasing staffing?
These are the types of operational outcomes Healthcare IT News identifies as the more meaningful measures of automation. healthcareitnews.com
At Sheba Medical Center, for example, deliverz.ai initially focused on automating chemotherapy medication delivery. The broader lesson from the deployment was that the value extended beyond an individual robot or workflow: the same operating intelligence could be applied to additional physical workflows across the hospital. healthcareitnews.com
Healthcare has already undergone one enormous digital transformation: digitizing patient information.
The next transformation may be about digitizing and intelligently coordinating hospital operations themselves.
That means moving away from separate automation islands toward an intelligent operational layer connecting robots, human transporters, equipment, infrastructure, and hospital systems.
Over time, repetitive physical tasks that can be automated increasingly will be. Tasks requiring people will remain human - but those people can be coordinated through the same operational intelligence.
The result is not a hospital run by robots. It is a hospital in which people, autonomous systems, and infrastructure operate as one coordinated system.
That is the larger promise of Physical AI: turning intelligence into action and making the physical hospital as connected and intelligent as the digital one.
Physical AI applies artificial intelligence to systems that interact with and perform tasks in the physical world, including autonomous robots, intelligent infrastructure, and operational orchestration systems.
Generative AI primarily creates and processes information, while Physical AI connects intelligence with physical action - helping determine what needs to happen, which resource should perform it, and how physical operations should be coordinated. healthcareitnews.com
Yes. Autonomous hospital robots are one application of Physical AI, but the broader concept includes the intelligence that coordinates robots with people, equipment, infrastructure, and hospital workflows.
Physical AI can automate repetitive physical work, dynamically prioritize tasks, coordinate resources, manage service levels, identify bottlenecks, and ultimately help hospital operations become more efficient and predictable.
Physical AI can automate repetitive physical tasks while allowing healthcare workers to concentrate on activities requiring clinical expertise, judgment, and human interaction. In hybrid operations, AI can also coordinate human transporters alongside autonomous resources.
