Physician-controlled movement is the step we are working on now. The step after it is where our products, if successful, could change how GI medicine is practiced.
We are working toward robots that could carry out a complete examination of the stomach on their own including visiting every region in a repeatable order, holding each view long enough to capture it well, keeping track of where they have been, documenting any abnormalities observed, and handing a physician an organized, complete picture of the stomach. Nobody’s hand on the controls. Highly scalable. That is the goal.
What we mean by autonomy, and what we don’t
When we talk about autonomy, we do not mean a robot that makes medical decisions. We mean a robot that does the coverage work. It can systematically get a good look at the stomach. The judgment about what that look means, and what to do about it, stays with the physician. Autonomy, done right, would move a physician’s attention away from driving and toward deciding.
Why it would matter
If we reach that goal, a few things could change at once.
The first is standardization. What we are chasing is an examination that can be defined in advance and held to a standard: the same regions, in the same order, every time, with coverage that is measured rather than assumed. Medicine has always valued procedures like that, because a procedure you can define is a procedure you can audit, teach and improve. Whether a small robot can deliver one inside a moving stomach is exactly the kind of question we will have to answer with evidence.
The second is reach. Today, an examination of the stomach happens where the specialist is. A system that could complete the examination itself, with the physician reviewing rather than steering, might not need that expertise in the room. Where the physician would need to be, and what it would take to do that responsibly, are open questions, not settled ones. But if we get there, it could change where our platform could be used: potentially far beyond the settings that have specialist expertise today.
The third is time. The mechanics of an examination take a clinician’s hands and attention. If a machine could take on that part, some of those minutes might go back to patients and to decisions. How many, and whether that holds up in real settings, is something we will have to measure.
The fourth is data. A robot that followed a standard path would produce valuable information which is what a learning system needs. It is the engine behind everything else.
Why it is hard
We would be doing you a disservice if we made this sound close. Here is what stands between us and it.
The stomach is not a still room. It moves, folds, fills and empties, and the only light comes from the robot itself. A robot trying to map the space has to do it while the space changes shape. That is a perception problem we are only beginning to understand.
Control is unforgiving at this scale. Holding a position in fluid, turning to a precise angle and moving deliberately, all on a battery the size of a baby’s fingernail, is the kind of problem that looks solved on a bench and turns out not to be in a moving organ. Every improvement in motors, power and control buys a little more capability, and there is no shortcut.
Completeness is surprisingly hard to define. How does a robot know it has seen everything? A physician answers that with training and experience. A machine needs the answer written down, measurable and testable. Working out what a complete examination means, precisely enough for a robot to be held to it, is part of the work.
A robot must know its own limits—and that must be built in from the start. An autonomous system needs to know when to stop, when it is uncertain and when to hand control back to a person. The physician-in-the-loop path we are building now is not something autonomy replaces. It is the foundation autonomy has to rest on. And autonomy is not something we will announce, it is something we will have to demonstrate with data, step by step, to regulators and to physicians.
Not one robot. Many.
Looking further out, we do not imagine a single robot that does everything. We imagine several, each built for a different job, and eventually working together: one that maps, one that inspects, and, much further along, ones we imagine doing more than looking. A coordinated group of small robots, each doing what it does best, is a whiteboard idea today. We mention it because it shapes decisions we are making now about how we design our systems to see, move and share information.
Earning it
Autonomy is not about removing the physician. It is about putting a physician’s judgment where it matters most and letting the rest of the work happen wherever a capsule could one day go.
The path runs through everything we have written about so far, from seeing and steering to knowing where you are and fitting it all into something small enough to swallow. The work we are doing is not easy, but we are up for the challenge. We look forward to updating you on our progress.
Arman Nadershahi
CEO
Endiatx
