Mohammad Noshad’s Shyld AI is Turning Hospital Disinfection into an Autonomous System
As hospitals look for more efficient ways to manage disinfection, this co-founder is exploring how autonomous systems can bring physical AI into healthcare workflows.
Mohammad Noshad, co-founder and CEO of Shyld AI, has spent the past several years building autonomous systems that disinfect hospital rooms without waiting for a person to do it. His work points to a broader shift in where artificial intelligence (AI) is heading. AI can be good at providing information, identifying patterns, and making recommendations. The next step is getting AI to act in the physical world.
For founders who are building AI in regulated industries, that distinction matters. The opportunity is not simply to make existing workflows more intelligent. It is to identify where human intervention creates friction, determine whether automation can safely close that gap, and build systems that can translate intelligence into action.
That is the emerging promise of physical AI. But robotics alone is not enough, Noshad says. A robot can perform a task, yet the larger challenge is using AI to coordinate that physical capability within the environment where the work needs to happen. For Noshad, that is where Shyld AI fits, using AI to make physical automation more responsive, more targeted, and easier to integrate into hospital workflows.
The cost hospitals cannot shake
Hospitals have made progress against infections that spread inside their own walls. According to a CDC report published in July 2026, roughly one in 38 hospitalized patients now has a healthcare-associated infection, down from one in 31 in 2015. The same report found that two of the categories that matter most, surgical site infections and pneumonia, have barely moved.
The infections that remain still carry high cost. Forbes reported that a single central line bloodstream infection costs a hospital roughly $45,814 to treat, and that infections across all categories add up to between $28 billion and $45 billion a year nationwide. Medicare has withheld the added cost of treating a number of infections that patients pick up after admission since 2008, and hospitals in the worst-performing quartile still lose an additional 1% of their Medicare payments under that penalty program, according to CMS’s current fiscal year guidance.
For Shyld AI, that made the problem less about whether hospitals needed better information and more about whether the systems already in place could consistently turn that information into action.
Why manual disinfection falls short
Most hospitals already use some form of ultraviolet disinfection. Noshad has estimated that 70% to 80% of hospitals have adopted UV-C technology, though that figure is his own estimate and has not been independently verified. He described the typical setup as manual, labor-intensive, and inconsistent. The tools built to flag problems encounter a similar limitation. Noshad also says that whenever a person has to carry out a cleaning step by hand, the process stays prone to error, a pattern he has pointed to in interviews about why Shyld AI builds hardware that acts on its own, closing the loop without a dashboard in between.
This points to a broader shift in automation. Information is useful only up to the point where someone still has to translate it into a physical action. Physical AI attempts to close that gap, allowing systems to perceive an environment, make decisions, and execute tasks within it.
But the transition is not simply about putting more robots into hospitals. Robots still need to operate within complex environments, alongside people and existing processes. The harder problem is making those machines work intelligently enough to become part of the workflow rather than another system that staff have to manage.
Building AI that finishes the job
Shyld AI’s answer is a wall-mounted device that pairs onboard sensors with targeted UV-C light. According to Shyld AI, the sensors continuously scan a room, assign a contamination risk score to high-touch surfaces, and then trigger a calibrated UV-C dose without a person starting the cycle. The system runs on VERTEX, the company’s own foundation model, processing environmental data locally on each device, rather than relying on a hospital’s cloud infrastructure for that processing.
A peer-reviewed study conducted at Stanford Hospital’s Advanced Endoscopy Unit, co-authored by Noshad alongside Stanford researchers Monique T. Barakat and Timothy Angelotti and published in the American Journal of Infection Control, found that the system reduced cumulative microbial bioburden by more than 93% compared with a control room running standard manual disinfection.
For Noshad, the significance of the result is not simply the number. It gives hospitals a measurable way to evaluate whether the system is doing what it was built to do.
Where automation meets hospital economics
Cost may be one factor drawing hospitals to the technology. Noshad has said, though the figure has not been independently verified, that it saves hospitals between $2,000 and $4,000 per operating room in monthly labor costs. Shyld AI also sells the system on a subscription basis rather than as a capital purchase, which keeps the decision out of the budget cycle that usually slows hospital technology adoption.
That structure also shapes how the company positions the technology to staff. Environmental services teams keep doing their jobs. According to Shyld AI, the device is designed to handle the repetitive disinfection cycles staff would otherwise perform by hand, freeing them to focus on work that still requires human judgment.
Lessons for other health tech founders
The broader lesson is that AI is moving from providing information to acting in the physical world. Robotics can perform a task, but AI can determine when, where, and how that task should happen. That coordination layer is what can turn individual robotic capabilities into practical automation.
Shyld AI’s next test is scale, with Noshad saying the company plans to extend the approach to operating rooms and pharmaceutical cleanrooms. The opportunity is not simply to build machines that can work, but intelligent systems that can help those machines work effectively alongside the people and processes already in place.
This article is intended for informational purposes and does not constitute medical advice.
Entrepreneur Media was not involved in the creation of this content.