In the operating theatre, the surgeon is the most visible part of the process. But a large share of the real challenge sits behind the scenes, in scheduling, logistics, and resource planning, where hospitals still rely too heavily on subjective judgement.
For years, digital health has often been associated with telepresence tools, including remote observation of procedures in the operating room. That phase is giving way to a more intelligence-driven model of surgery.
Rather than treating the operating theatre as a simple workflow, new systems are approaching it as a data problem that can be analysed at scale. Using computer vision for measurement and generative AI for predictive scheduling, Proximie is turning the operating theatre into a more structured, data-rich environment.
The company says it handles around 120TB of unstructured video data through a hybrid edge-to-cloud setup, with technical controls designed to protect patient data sovereignty. Its long-term goal is to make AI an invisible layer in the hospital environment that reduces manual effort without becoming a standalone tool.
The ‘hanging around’ problem
Proximie’s starting point is not the surgery itself, but the global shortage of safe surgical access. Around five billion people still lack access to safe procedures.
The company’s CTO argues that building new hospitals and training staff takes too long, so the better approach is to improve the use of existing resources. The main issue is not making surgeons work faster, but reducing wasted time between steps.
To address that, Proximie uses ceiling-mounted sensors to create a precise record of operating room workflows. Unlike human memory, which can be incomplete or subjective, computer vision can track exactly when a patient enters the anaesthetic room and when they leave the procedure room. That makes it easier for hospitals to identify where dead time is happening.
The predictive scheduler
This data helps Proximie tackle one of the hardest parts of hospital management: elective list scheduling. If a case is underestimated, the full day can fall behind. If it is overestimated, operating room capacity is wasted.
By analysing three years of electronic health record data, Proximie’s AI can now outperform human schedulers in some cases. It looks at variables that are difficult to calculate manually, including the relationship between a patient’s BMI, age, and the specific surgeon-anaesthetist combination.
The impact can be immediate. At St Thomas’ in London, thoracic surgeons have added one extra major case per day by using real-time data to tighten scheduling.
From ‘Zoom for surgeons’ to unbounded AI
During the pandemic, Proximie was often compared to a video platform for surgeons. That telepresence function helped normalise digital access to the operating room, but it is no longer the core product.
The harder problem is the unstructured nature of video and audio data, which has historically been difficult to process at scale. In healthcare, the number of possible variables is effectively endless because every patient is different.
The company sees healthcare as an unbounded problem, and believes current AI systems are finally capable of reasoning across that complexity. The aim is for AI to become part of the background infrastructure in hospitals, taking away routine work rather than acting as a separate gadget.
Technical architecture – edge vs cloud
Handling 120TB of unstructured data across multiple regions requires a hybrid architecture that balances latency and privacy.
Edge devices mounted on operating room ceilings handle privacy at the source. They obfuscate and redact sensitive information before any data leaves the room. The company says no unmasked data ever exits the operating theatre. After redaction, the data moves to the AWS cloud for large-scale asynchronous processing.
For real-time frame-by-frame analysis, edge computing becomes essential. In time-sensitive procedures, the system may have only 18 milliseconds to analyse a frame at 60fps, which leaves no room for cloud latency.
The encryption moat
When operating across different jurisdictions, data sovereignty is a strict requirement. Proximie uses AWS Global Accelerator to route and store data within the correct jurisdiction. The workflow, rather than the user, determines where the data goes.
On concerns about the US Cloud Act, the company points to its encryption model as the main defence.
Even if legal access were requested, the data would remain unreadable in practice because the encrypted form is not directly usable by outside parties.
Safeguarding against hallucinations
In a real clinical setting, AI hallucinations cannot be tolerated. Proximie addresses this with a human-in-the-loop governance model.
The AI can suggest a “win of the day” or highlight the biggest efficiency opportunity, but it cannot make executive decisions. It must also explain its reasoning and show the data points used to reach each recommendation.
That traceability allows theatre managers and clinicians to review the output and keep final control over the workflow.
The future of surgical logistics
As Proximie scales, its focus is on making AI even more of a background utility. The company is targeting the infrastructure and logistics problems that human teams still struggle to solve consistently.
The shift from telepresence to intelligence-first operating systems is central to its strategy. By combining cloud-scale processing with edge-level privacy controls, the operating room is moving toward a more objective, data-driven model of efficiency.
