Welcome to Consensus, a podcast from Census Technologies. Hey. What's going on, y'all? It's Daniel Littwin, voice of B2B. If you're tapping into this episode, guess what? You're tapping into a part two. You may have missed part one. So make sure you're heading back to part one of this conversation with Harshil and Seamus as we discuss AI and sterile processing departments. If you did catch part one, well, you're in luck. Here's the rest of the conversation. We're gonna pick back up where we left off as we continue to dig into practical use cases for AI and sterile processing departments as we unpack the consensus final check solution that is bringing AI to SPDs, And we look ahead to how AI is gonna continue to shape SPD workflows, culture, and more. So let's go ahead and dive back in with Seamus and Harshal. Harshal, I wanted to pick your brain here on, you know, AI in action in context of the real workflows in an SPD. So there can often be tool fatigue in really any work environment. This is not unique to SPDs. But you roll out a new tool. Yes, it's useful. Yes, it improves workflows, but man, I gotta learn all the nuances of it, or I gotta change how I approach my work to make sure I take advantage of the tool. There can be some friction there around continued education, L and D, regardless. So I wanted to get your thoughts on how did y'all approach your tool to integrate AI in such a way that it doesn't slow down teams, doesn't add complexity, right, and really becomes just an intuitive addition to all the great resources they already have. What are your thoughts there? Yeah. So so we we thought about this a lot. And I I think one of the key objectives for us, right, as we went through the final check was to make this super simple to use, right? It, we, Our goal was that we wouldn't want the technicians to stop what they are doing, open up another application, open up another browser, go and launch FinalCheck, validate it there and come back and then go and execute your workflow, right? So that would be disruptive to their standard work. So we made sure that this is embedded into their workflow. It's very simple to use. It's right there and there as they're doing assembly, right? They can just open up final check. It does a quick validation. You have the integrator present and boom, you're done, right? And it's part of the technician's standard work, right. So very, very simple to use. And that was our key goal, right. So it doesn't become another set of tools which they would have to learn, but this is something which is right embedded into that process. And the other thing, you know, we kind of touched upon that earlier about the traditional software tools and AI, right? In this scenario, because of so many different variations, it was also important to make sure the AI computer vision model which we have built for final check also learns as it encounters new scenarios, right? So we call that as if it identifies if an object exists it's object detection and then it also needs to know what is that object right or what it does and that's object recognition and our solution does that so no matter you know where you put that integrator in whatever lighting conditions it is able to identify that right so it actually learned as it encountered new scenarios so you know and you know as Seamus already shared right we are seeing a lot of impact right with those number of cases where you don't have an integrator really dropping down to almost zero Right? Which is directly driving the key goals of SPDs. Right? To deliver high quality trays at the right time to support the OR. So let's go back to some of the impact storytelling here. Seamus, would love to hear from you here. You already gave us an example a little earlier, but I'd love to just expand on it a little more. If you've got any other anecdotes in your belt that we can pull from here. Can you share an example of how final check has changed outcomes for a customer? Know, if you could even paint the picture of a before and after situation here and would love to hear what kind of feedback you've heard from technicians or leaders that are using the tool. Good point. So yeah, a couple of examples already shared where customers had dozens of trays going up the OR brought that down to zero. That was That's pretty crazy. It is pretty crazy. And it was kind of amazing to see, especially since so many other things had been tried and not had a lasting impact. But it's more than that too. So we talk a lot about how final check checks to make sure the integrator is in the tray. But in that process, it's also documenting assembly. It's taking a picture of that assembly process. So we had one customer who SPD manager, actually SPD director got a very urgent call one Monday morning. The OR director was very upset that two trays had been used in procedures that were clearly not sterilized. And, you know, that's a situation no manager SPD director wants to walk into. So, he was able to use the pictures captured by final check to see what had happened. You know, so he looked up, who assembled these trays? Let's see some pictures of these trays. And when he pulled them up, sure enough, they were clean as a whistle. They had been properly cleaned and properly sterilized. And because he had that, those pictures, because he had that data, the conversation changed. They started doing some problem solving and discovered that what had happened was the trays had been used and then put back on a clean shelf in the OR. And that's a crazy scenario, but every now and again it does happen. And so he's able to use this system not just to provide better trace to the OR, but also to document what's happening in sterile processing and to ensure it's, again, it's that extra layer of trust, right? So you take a relationship that might be antagonistic, might be a relationship where the default is always to blame SPD when something goes wrong because they're at the bottom of the totem pole. And now, you know, SPD manager can bring the receipts. So, you need to have your ducks in a row if you're going to try and blame SPD. And that changes things, right? But more importantly, it gives you data to help build that trust between the two departments. So yeah, that has been huge. In general, the feedback I've gotten from SPD leaders has been all around positive. They do, they like the fact that they can now know that trays are being sent up with the integrators in them, that you can run a report now to see, hey, what did we send up today? Show me that everything got checked, right? You could sleep at night. They love the fact that they have that image data. They can now and if SPD does make a mistake, they have the image, they can use that for training and education. They can use that to improve the department. That's something they didn't have before. Technicians, every now and again we talk to a technician that is concerned about whether or not it's being used to spy on them. But I know and I understand it. But we designed it so the technician is in complete control of that camera. They physically have it there. They can point it at anything they want. They can point it away from them if they need to and it has a light on it that comes on when it's recording. If the light's not on, it's not looking at anything. Right? And once technicians understand that this is documenting what they did right so that nobody can come back and say they did it wrong, they tend to really like it as well. Every technician has the same experience when your supervisor pulls you aside and says, hey, this tray that you did two weeks ago, you have an integrator in it. Well, what are you going to do? You know, you don't have any documentation to prove that you did it right. You know that you put those integrators in there every single time. So how could you possibly forgotten it? Right? And so you feel like you're getting blamed for something that maybe you didn't do. And now you have a picture to prove that you actually did it. And you have a tool that ensures that you're not going to make a mistake. And so once technicians realize that, the feedback tends to be very, very positive from them as well. All right, folks. I think we've gotten some good, you know, boots on the ground commentary here for how final check is already delivering results to SPDs. We've talked shop on how AI should even be thought about strategically as another layer of tools, another layer of insights to support SPD techs and leaders. So I wanna wrap things up by looking ahead now, right? At what's next for AI and sterile processing. Now that we already have some data, you guys are leading from the front here at Census with delivering great AI tools that improve SPD workflows. Now we get a sense for what is working, what still needs support and how to continue to drive these tools forward. So, Harshal, let's start with you here as our CTO perspective. When you look ahead at the industry at the development of AI as a tool set, where do you see AI having the biggest impact in SPDs over the next, I don't know, three to five years or something? What are your thoughts? You know, as I look at the impact of AI, right, I'm seeing the trends in the next three to five years move more from a reactive manual labor to a proactive automated orchestration using AI. And I'll touch upon like three big areas in this. Right? One is getting to that quality focus. How do we get to zero error assembly? And AI will have a big role to play there. Computer vision models will become the standard providing real time verification of every instrument to virtually eliminate those tray errors, eliminate bio burden risks before they reach the OR. So have complete high quality trays, zero errors delivered to the OR, right? So that's kind of the quality focus. The second is more predictability, more predictable operations, as we touched upon earlier in this conversation. AI will synchronize SPD throughput to OR demand real time, you know, prioritizing trays based on the surgical schedules, and also predicting those equipment failures right before they cause downtime. So it will help leaders make better staffing decisions and scheduling decisions. So that's the predictability focus using AI. And the third I would touch upon is a workforce multiplier focus. Essentially, we are already seeing, you know, robotic instruments, robotic arms being used for minimally invasive surgeries. Right? It's happening now in many countries. The same will flow into SPD, right? We should have automated robotic arms which will handle decontamination, you know, reduce the technician exposure to biohazards, and improve the ergonomic safety of SPD. So essentially it becomes a workforce multiplier. So I think those are some of the immediate trends, you know, essentially you know a technician's role I think would move more into this quality orchestrator, as I call it. Know, AI should be able to handle millions of these variations and whereas humans will continue to have that accountability. They would have to oversee the systems, manage those exceptions, but and really have that, you know, what humans do the best, right? Have that judgment to make sure that, you know, we are not compromising on patient safety. So that's how I see, you know, the next three to five years evolving, especially in the SPD space. Now Seamus, as a follow-up here, from a product and customer standpoint, what do you think good AI is gonna look like in an SPD moving forward? Right? Where should we continue to develop its use cases versus where might it be a little more experimental or not quite as well deployed? Yeah, I think so. It's a good question. We've thought a lot about that here, obviously. I think when you think about AI compared to traditional software, think about the kinds of problems that it is well suited to solve. In SPD, if every single instrument were the same, you know, and processed the same way, the same steps, you could just build a big machine that would process all your instruments. But they're not. There's hundreds of thousands of different instruments. There's lots of variation in how they have to be processed. There's lots of different nuances between instruments and how you have to check them for readiness, for cleanliness, for bioburden, all of these things. So, SPD technicians make a lot of little decisions all day long. And that is where AI can help. You know, when you think about, we talked about it earlier about IFUs, you know, about how they're laid out, they're not written for a technician. They're not there to help the technician process things correctly. They're there to help protect the manufacturer. AI can help bridge that gap. AI can ensure consistency. AI can augment the technician. Harshal touched on this a little bit, but it really is about augmenting the capabilities of the technicians at SPD. I don't think that time is going to go backwards. And we're going to go back to a time when staffing is not a challenge in SPD, because frankly, staffing's always been a challenge in SPD. It's just worse today than it's ever been. And so, you know, that economic pressure is going to continue to bear down on hospitals. It's going to continue to bear down on sterile processing departments. And so, I think good applications of AI are going to augment the technician. They're going to be force multipliers, as Harshal said. They're going to continue to drive productivity. I think honestly, one of the things Harshal said that needs more attention is this idea that we can get to zero defects to perfect trays. And that sounds crazy, but I think it might be possible. We're not going to get there without the assistance of AI. But with AI, I think we really could deliver on that promise. And that's kind of an amazing thing. Harshal, back to you real quick. You know, we mentioned this a little earlier. The question of AI always comes with the question of, you know, ethics, the responsibility of AI as a tool, whether that's privacy or the implications of leaning on the insights of a tool like AI for decision making, whatever. There's lots of different nuances and you know, reasons to raise a finger. But I wanted to get your thoughts on practically how should teams think about adopting AI responsibly, especially in SPDs? What are the nuances that they should consider versus the ones that maybe you're a bit overhyped or are more, you know, more anxiety inducing for the headlines than really for the practical reality of SPD operations? Yeah. It's a great question. I think, you know, that the big, the way SPD and leadership should think about AI is AI is a copilot, right, and not an autopilot. Essentially, you know, there always needs to be this human in the loop oversight because the final accountability is still with the key experts there, right, in the SPD. And so how can AI be considered as an assistant, a smart assistant and it actually empowers technicians to get to their goals quickly and efficiently right. So having that human in the loop oversight is critical. AI should be transparent, you know, to your point, you know, it should, we should be able to explain the outcomes of AI, right. If, you know, why there have been the results, right. Why was result X versus result Y. We should be able to explain those outcomes and that's a responsibility, you know, we take very seriously, right so the parameters which are used how the data is processed making sure that the data remains in the customer's tenant right it's not going outside to train these large language models it remains in customer's own environment and those are critical, right, to build that confidence. We work with a lot of the customers. Even yesterday Seamus and I were on a call with a customer and they all have these AI governance bodies now, right, which actually evaluated these different tools. And I think as you're able to explain how the AI functions and why the results are what they are, it just creates a lot of confidence with these governing bodies also. And the third I would state is just it needs to be incremental and measurable, right, which will create that confidence with the leadership. So start with narrow high impact use cases, right? For example, how do I have my tray being complete? How do I validate the tray accuracy? And prove that out in partnership with the customer, in partnership with the frontline technicians. And then you scale responsibly. Right? So that builds trust with the teams and with the leadership alike. So that's how I would say the effective and responsible use of AI will scale as that trust is built into the system. Alright, folks. So I'll leave it with some final questions to both of y'all. And, you know, feel free to kinda rapid fire here, answer the question to the other person as well. I know, I'd love to hear from both of you here to wrap things up. So, Harshal, we'll start with you here. What's one misconception about AI and SPD that you'd like to clear up for people as we as we close the door here? Give us your lightning round answer. AI doesn't replace people. It amplifies human capability. It acts as a second pair of eyes. I like that. Seamus, how would you answer that question? Maybe a little more wonky and technical. I think there's a lot of misconceptions, not just in SPE, but in everywhere about how AI actually works. I think a lot of people think AI is sort of a genius in a box. And that's sort of right, but not exactly right. In many cases, it's a very sophisticated interpolation machine. That is, you can give it a lot of dots and it can fill in the connections in a way that a human does. But there's things that humans do very well that AI doesn't do and that's extrapolation. And that's a subtle point, but it's an important point. So there's a, I think there's, it drives a lot of misconceptions about what AI is going to do or become over the next couple of years. It really understands like that that subtle difference underscores why Harshal said we should think about AI as a copilot, as a way to augment the technician, not as a way to replace the technician. Because again, there's things that humans do very well that AI isn't even designed to do. So you have to have that human in the loop. The human always has to be responsible ultimately for, what happens. Alright. Now, Seamus, I got a question to start with you. A rapid fire here, response. What should SPD leaders be thinking about today as they prepare for what's next? AI can be part of your answer or not. What's your lightning round response? Yeah, I think right now, one of the things they should be thinking about is the way in which the regulatory environment might change. I know everybody's already thinking about this, but what I mean specifically is not just the regulations and how they might change, but how those are enforced. There are some changes going on right now, you know, at JCO360 and things like that, that I think are going to drive some changes in the way we run sterile processing departments. And I think ultimately, it's going to be for the better. But I think that that's something that should be top of mind for everybody. And again, that's another opportunity for AI to really help out. That's going to be an important lever for SPD leaders tomorrow. And Harshal, same question your way. I think the SPD leaders and the leadership at the hospital, right, in general should think about start with the use cases first. Start with the business context. What are we trying to solve here? Right, AI is an enabler. And there could be you know depending on that specific use case. Maybe AI is the answer, maybe AI is not the answer in certain other situations. Right? So I would suggest instead of saying, oh, whether AI will fit into this or not, it should be the other way around. What's the business problem we are trying to solve? Start with that, right? What's that high impact use cases which is going to improve people's lives, drive patient safety, improve the technician's life in SPD, right, help them to do their jobs better. And then we can figure out, you know, from a tools perspective and solution perspective, you know, what should be the right technology. So that's how I would start with. I mean, you would hear about a lot of things, you know, we are hearing about agentic AI and the robotics, right, which we talked about. All of those have strong applicability, but it just depends on the specific use cases which we are trying to solve. So that's what I would suggest, you know, start with the business problem, and then we'll look at the technology, how it fits. Alright. You two. I think with that, we'll go ahead and put a pin in this episode. Thank you, Seamus. Thank you, Harshal, for your great insights today. This has been one of the most practical episodes we've done in a while. I've loved hearing all of the impact stories. I mean, going from, you know, dozen plus tray errors to zero after implementing this platform is, I think, a testament in and of itself that AI can be launched responsibly. It can be launched intentionally to actually affect SPD workflows in a positive way that helps address some of the underlying tension, anxiety, frustration that some of the SPD techs and leaders can face boots on the ground here today. So thank you to the two of you for all these insights, and I'm really looking forward to seeing how you guys continue to innovate for sterile processing departments, how you keep putting AI to work as a tool that augments the human element of health care. Again, folks, we've been chatting today with Seamus Johnson, who is senior director of innovation at Census, as well as Harshal Guradia, chief technology officer at Census. Harshal, if folks wanna get in touch with you, how can they learn more about you, your role, and maybe shoot you a ping? Yeah. Absolutely. You know, would love to connect. Please feel free to reach out. You know, I'm very active on LinkedIn. Right? So please feel free to reach out through that. Right? And again, if you want to interact you know we have these consensus channels right so you can reach out through that And, yeah, I think if you want to have a chat, if you have to have a conversation, would be would be happy to have that. Seamus, same question your way. Folks wanna get in touch. They wanna learn more about you and your insights. How can they reach you? Yeah. So yeah, you can hit me up on LinkedIn. That's easy enough. But honestly, everybody here at Census knows who I am. If you talk to your client manager, if you talk to your service desk, just say, Hey, I want to talk to Seamus. They'll put you in touch. Easy peasy lemon squeezy. You said it yourself. Most popular guy at Census. Don't know. Nobody knows who I am. I don't know. All right, right. I love it. Well, whatever. We'll give you the popularity badge regardless. Alright. Harshal, Seamus, thank you to the two of you for your, perspectives and, all your input today, and I'm looking forward to chatting again soon. Thanks again. And thank you everyone for joining us on this episode of the consensus podcast. Again, a Census Technologies podcast. Folks, as we wrap up today's conversation, you know, one thing really stands out to me, I'm sure you, already have this on the mind that you're taking it away as our audience. AI and sterile processing is is not about replacing people. Right? It's about supporting teams at the moments that matter the most and providing that extra layer of oversight and support. So whether it's reducing rework, whether it's cash catching miss indicators, building confidence at the end of assembly. Right? What we discussed today shows how AI can be applied thoughtfully, responsibly, and again in ways that truly fit SPD workflows. Make sure you're heading to our website, census dot com for more. To learn again about Assembly Copilot final check, make sure you're subscribing to Consensus on Apple Podcasts and Spotify. You can find more episodes on our website as well. And we're looking forward to more great conversations with the innovation technology team here at Census. Alright, folks. We'll catch you on the next one. I'm Daniel Littwin, the voice of b two b. See you on the next episode of the consensus podcast.
