AI and Healthcare: Adrian Jennings, Chief Product Officer at Cognosos, on Scaling RTLS in the Post-AI Era

On any given day inside a hospital, thousands of micro-decisions unfold quietly. A nurse searches for a clean infusion pump. A transport team moves equipment between departments. A safety alert triggers in a corridor. None of it makes headlines. Yet each moment carries operational weight.

For more than 25 years, Adrian Jennings has been thinking about one deceptively simple question: How do you know what is actually happening inside complex physical environments in real time?

As Chief Product Officer at Cognosos, Jennings works at the intersection of AI and healthcare, cloud architecture, and real-time location systems. His focus is not on speculative AI or futuristic diagnostics, but on something more foundational. He calls it ground truth. Where are the assets? Where are the people? What is interacting with what? And how can that visibility translate into measurable operational impact?

Cognosos entered the healthcare technology market later than many of its competitors. That timing, Jennings argues, turned out to be an advantage. Built in the era of cloud-native infrastructure and AI-first engineering, the company set out to resolve a long-standing tension in real-time location services: high performance traditionally meant high cost.

Low cost meant low accuracy. The middle ground remained elusive.

You have spent more than two decades in real-time location systems. For CTOs evaluating infrastructure modernization, how do you define your role as Chief Product Officer, and how do you align long-horizon product strategy with corporate objectives in a rapidly shifting AI landscape?

Adrian: My role is Chief Product Officer. So, the Chief Product Officer means I oversee and define the product strategy and product roadmap. And that has to fit within the context of our corporate goals.

But the goal is always to stay one or two steps ahead of the market so that we make sure that we’re engineering for the future. So that’s the role. My journey here was a long one. I’ve been in this business, this real-time location business, for just over 25 years now, and at Cognosos only for five.

So the first 20 years, I was focused on super high-end, super-accurate location technologies that did very difficult, esoteric, and frankly, very expensive things, while the market really was crying out for something a little bit different. The market was split into very high-cost, high-performing systems and very low-cost, low-performing systems. But nobody had cracked the high-performing, low-cost model where the real magic is.

And that’s what brought me to Cognosos. Cognosos actually came a little bit late to this RTLS business, and in doing so came and reinvented the whole thing. Cognosos came in the era of the cloud and AI and applied AI in ways that nobody had thought to do before and actually, and we’ll get into this, I’m sure, but cracked the code on how you marry high performance with low cost, which is really a massive accelerator to these technologies across our focus, which is largely in healthcare, but we focus in manufacturing and logistics as well, those two verticals.

When you talk about high performance and low cost, you are describing a tension that has shaped infrastructure strategy for decades. Can you ground that in a real-world healthcare scenario and explain where that trade-off shows up operationally?

Adrian: Yeah, so just to back up a bit and maybe think a little bit about healthcare and where some of the use cases are.

The biggest, highest-value use cases in healthcare are around workflow optimization on the one hand and safety on the other.

Workflow optimization means understanding the location and movement of assets, patients, and staff in real time and historically. Your historical records are the engine for process improvement, and your real-time status, we call it ground truth, what’s currently happening right now in the organization, that’s the basis for high-performing operations, reducing errors, increasing throughput, and so forth.

Then the other side of that is safety. That ranges from literal staff safety, making sure that nurses are safe. Unfortunately, violence against nurses increased dramatically during COVID and is continuing to increase right now. So part of safety is making sure that they have a mobile panic alarm, they can call for help, and we can orchestrate the help.

The other part is patient safety as well, making sure that hand hygiene compliance is kept up, for example.

Both workflow optimization and safety are predicated on a very high-quality understanding of what’s happening. That means where everything is, where it’s been, where it’s going, what is interacting with what, and who with whom.

The early stages of this technology were fantastically complex, difficult to install, difficult to own, and expensive to install and maintain. That’s the high-cost, high-end model. It worked, but the return on investment was never there because it was cost-prohibitive.

Then Bluetooth came along with the promise of very low cost. It’s commoditized and inexpensive, but it’s not very accurate. So the quality wasn’t there.

What Cognosos did was throw Agentic AI at the problem of how you use a Bluetooth technology, which is inherently not very good for location, and use the power of AI with a giant cloud-sized brain to do things that had never been done before and get the performance level of those higher-end solutions at a fraction of the cost.

So we’re AI down at the core, creating this ground truth of what’s happening. That allows you to pile AI on at the higher level to do things like orchestration, decision support, and process optimization. But nothing in that orchestration layer is worth anything if the ground truth isn’t good enough. And ground truth isn’t worth anything if you can’t afford it.

So high-quality, affordable ground truth is the foundation for high-quality operational inferences higher up the stack. That’s really what we’re all about.

AI systems are often described as probabilistic rather than deterministic. In a regulated environment like healthcare, how should leaders rethink governance and accountability when deploying AI-driven operational systems?

Adrian: Governance is really important. One of the things we do when engaging with a hospital is set up a governance committee because governance is very cross-functional.

The CNO has a stake, the COO has a stake, the CIO has a stake, the CRO has a stake. It’s very cross-functional. So the governance committee is really important to drive adoption and make sure they’re taking a holistic view.

Now, on the point about AI outcomes being probabilistic, I would dispute that. If your AI outcomes are probabilistic, you haven’t done a good enough job tying down the problem you’re trying to solve and the endpoint you’re trying to get to.

There’s a lot of AI under the hood in our platform, but the outcomes are measurable and predictable.

For example, one healthcare facility using us to manage asset workflows ended up with 100 percent availability of clean equipment, 100 percent preventive maintenance completion, a 92 percent reduction in lost and stolen equipment, and a 20 percent reduction in total rental days.

They were about to buy 200 infusion pumps and canceled the order because they didn’t need them.

Each of those metrics is concrete, measurable, and meaningful as KPIs in the organization. AI may be statistical in nature, but the outcomes absolutely can be measurable, should be measurable, and are measurable.

As agentic AI systems become more autonomous, do you see a structural shift in data ownership or accountability, particularly in operational healthcare AI versus clinical AI?

Adrian: It’s worth splitting healthcare AI into at least two categories.

The category we don’t play in is clinical AI, the diagnostic side of AI tooling in healthcare. That requires a lot of thought about adoption and everything you just asked about.

Where we play with Agentic AI is in the operation of the facility.

Am I correctly staging cleaned equipment? Am I correctly staffed for the census that we’re predicting? Are people being kept safe? Are we following processes and procedures?

I think AI is done well if there is specifically no requirement for anyone to change anything. We see what we do as very much supporting people and helping them do their jobs better, not creating new categories of work that maybe require new skills and new team structures.

In a perfect world, our AI slides into the background because everything just works when it’s running, and people forget how bad it used to be when things were broken and processes weren’t efficient and nothing connected at the right place at the right time.

So, I think a hallmark of good operational AI is one that specifically doesn’t require new ways of working but supports the workflows that exist.

Many enterprises struggle not with AI experimentation but with scaling and achieving measurable ROI. What practical steps should enterprise IT leaders take to ensure AI initiatives scale both technically and economically?

Adrian: When everybody says scalability, they usually mean scaling very big. But in healthcare, you need the ability to scale small as well.

There are very large IDNs with very large hospitals on large complex campuses. There are also many small medical office buildings, rural critical care hospitals, imaging centers, surgery centers.

So scalability is something that we take very seriously. Our solution scales very big, but it scales small too, even to the point of scaling out to home healthcare. And which is now obviously an increasing trend.

How do you get started? You pick a system that can scale small and scale big because that allows you to do pilots. Pick a department or a building and start piloting.

The trick is picking the solutions that will provide value at that small scale and give you the ability to grow both sideways into other departments and vertically in terms of use cases and depth of analytics. It’s important that scalability is future proofed. There have been too many adoptions in the past of technologies that were already reaching their technological cap.

AI at the core and AI in the platform is future-proofed in a rather unprecedented way. We all agree that AI is here to stay and is just growing rapidly.

Nothing makes it easier to expand a solution than a successful pilot that a governance committee can look at. And say it does what we were promised, it gives measurable outcomes, and we should adopt it everywhere.

Want to know more about AI and healthcare? Explore more insights.