AI in Clinical Trials: What You Need to Know

Michael Krigsman: Clinical trials are fundamental to developing new medications, but they face significant challenges, including lengthy timelines, high costs, and complex data management. I'm Michael Krigsman, and today on CXOTalk episode 877, we're discussing the impact of AI on this critical part of our healthcare system.

Our guest is my old friend, Jay Ferro, Executive Vice President and Chief Information Technology and Product Officer at Clario, a leading provider of clinical trial technology solutions.

Jay Ferro: I oversee IT, product, data, security, marketing, and corporate communications. All the things that you normally would find in the traditional CIO role. All the things that you normally would find in the CTO role, enterprise architecture, et cetera, and then product and marketing.

We also get to build the solutions that empower clinical trials all around the world, and I get to have a hand, although I leave it to the experts, in the marketing of those products all around the world for our customers, and ultimately, the patients that we support.

Unified Roles and Seamless Experiences

Michael Krigsman: Jay, it's a very unusual role, again, chief information technology and product officer. How does a role like that actually come together? I have not heard of anybody with that exact title before.

Jay Ferro: To me, it comes down to trust and delivery and transparency. I work for a terrific CEO. My team and I deliver. We're transparent. We own issues. We're business-focused first, technology-focused second.

In no way am I minimizing the importance of staying ahead of technology trends and leveraging emerging and transformational technology. That is not the point. But the point of an organization, especially with what we do, is serving our customers. And it's not just about the technology. It's about empowering clinical trials. Walking a day in the life of what a patient goes through, of what a clinical trial investigator goes through, a trial site, what our sponsors deal with when they're the ones funding a trial.

Truly understanding the business of what we do, I think has built trust with my CEO that I could have handled a broader role.

We have talked for years about understanding the language of the business, yet I still see so many CIOs, CTOs struggling with how to do that. They lead with technology and then maybe put a business wrapper on it. And it really starts with an understanding of the value chain: acquiring customers, delivering your solution, whatever that may be, or your product, servicing the customers at the end, making sure that they're having an amazing experience.

You're staying ahead in the innovation cycle so you're constantly delivering. And all of that stuff is powered by technology, but it's not technology that's the end game. It's removing friction. It's delivering value. It's delivering services on time. It's making sure things are up, of course. It's delivering high-quality data when you say you're going to do it. It's the new endpoints that are being brought into a clinical trial.

And when you can do that consistently and show that you have a good understanding of whatever domain you're in and you're having open and transparent conversations with your CEO about your aspirations... Part of this came from me just raising my hand and saying, "I'd like to... When you do have other opportunities, I'd like to be considered." And part of that has come through delivery. Part of it has just come through transparency and good relationships.

Michael Krigsman: What is the benefit or the value of combining these multiple functions under this particular umbrella?

Jay Ferro: Well, my CEO only has one throat to choke. He knows where to go if something's not on time or it's delivered with excessive bugs or defects.

But all kidding aside, we own the solutions, the technology portions of the solutions, all the way from inception, from a product management point of view, all the way to marketing. Now, we have other groups that are responsible for delivering our services, our solutions, our science, all of those things, and they're amazing at what they do. That doesn't fall under my remit.

But the other advantage it gives, Michael, is we're in a lot of different areas. We deal a lot with cardiac safety, with respiratory, with medical imaging, what's called eCOA, or electronic clinical outcome assessment. All of those things are different, but there are common threads.

Being able to unify those technology solutions or those products and those solutions into one tech stack and provide a more seamless experience for our customers... And our customers are, of course, the sponsors who are underwriting the clinical trial. That could be large pharma, perhaps a CRO or a biotech, the clinical trial sites all around the world who are actively running the trials on behalf of the sponsors, and then, of course, most importantly are the patients, the people enrolled in a clinical trial, wherever they may be around the world.

And we want to make sure they have the most frictionless, seamless experience they have. They're all overworked. They may be going through tough times if they're going through a clinical trial. We want to keep them front and center.

And because this role is now unified and we have one throat to choke and we have a unified product management organization, we were able to see and leverage technology and products all across our ecosystem versus having siloed products or siloed technology. And we now talk about Clario in terms of an overall ecosystem and platform versus a bunch of point solutions. And our customers want that. They want a Clario to deal with, not a Clario, comma, X, Y, Z solution ad infinitum.

Michael Krigsman: The ultimate point then, correct me if I'm wrong, is by unifying these teams across what are typically silos, you're enabling a hopefully seamless experience for your customers.

Jay Ferro: I think that's one of the largest benefits, Michael. I think one of the other benefits is internally, if you want to put a tech spin on it, is I'm able... And I say I'm as a proxy for my team, is able to see across our entire tech landscape, look for commonalities, look for differences. You don't want to build six of the same thing because you have product silos all over the organization that are doing their own thing.

Unifying that vision and being able to see where you can build once and deploy many is huge. There are the economic benefits of that. I build it once and I get to use it all over the organization. That's wonderful. It keeps my CapEx lower than it otherwise would be, keeps my spend, supportability gets better, all of those things that we care about.

But at the end of the day, the downstream impact is going to be felt by the customers, the sites and the sponsors. They're not getting a different experience every time they interact with a different product line or anything. They're used to seeing things a certain way, they're used to dealing with Clario a certain way, and that is really what we're striving for.

If you've ever walked into a clinical trial site, and I'll describe it for you... And these investigators are working their tails off, they're terrific, hardworking, brilliant folks. You see stacks of devices, laptops, respiratory devices, cardiac devices, and stickies on each one of them to talk about which company it's from, which study it's in.

All very important to keep everything organized, but I can't control what my competitors do. I can't control what other parts of the clinical trial ecosystem do. I can control what Clario does! And what I really want more than anything is when an investigator or a patient deals with Clario, whether it's a BYOD app during the life of a trial, a connected device during the trial, a portal that a sponsor is going into, is they have a frictionless, seamless experience.

And if that sounds familiar, it should, because really, every industry wants that. And every industry I think thinks it's unique in that, "Well, we can't possibly think about it in consumer terms," and I would argue that you absolutely have to think about it in consumer terms.

If every time I went to Bank of America and I had 17 different apps, "Well, this isn't the transfer app. You got to use that. This isn't the loan app. That's a different app. This isn't the check for your balance app. You got to fire up something different," you'd be like, "Man, I'm out of here." And we really want that unified experience while retaining speed and quality, integrity of the data.

Michael Krigsman: Very, very interesting as you're describing the comparison of this clinical medical system. But from a user experience standpoint, we're all just people and we want that non-siloed, seamless experience, as you were describing with a bank.

Jay Ferro: You nailed it. And that's it. That's it. And I think the secret is you always have to... We are in a highly regulated industry, so in no way am I saying that you freewheel it just to make it ease of use and that's your only bar. There are regulatory hurdles. There is scientific validity that you have to meet, and we are always focused on the science.

My chief medical officer, Dr. Todd Rudow, and I have a great relationship, and the hundreds of scientists that we have in our organization are amazing partners. But within the context of keeping everything valid and meeting our regulatory burden, as much as I can, I want to make that a consumer-like experience and as friction-free as I can.

Customer Concerns and Future Challenges

Michael Krigsman: And I just want to tell people that you should ask questions. If you're on Twitter, use the hashtag #cxotalk. If you're watching on LinkedIn, just pop your questions into the chat. And truly, it's a unique opportunity to ask Jay Ferro pretty much whatever you want, I hope you folks take advantage of it.

We have, let's see, our first question comes in from Chris Peterson, and he says, "This is a touchy subject. In Clario's niche, are you seeing or expecting customers hedging because of research funding cuts?"

Jay Ferro: We're certainly keeping an eye on that. And you can't help but read everything that's going on in the news and wondering what the knock-on effect is. And thank you, Chris, my good friend, Chris, who has been a regular viewer of yours. I know that. And I am going to say that the last time I was on here, which was years ago, he asked a question, so I am honored that he showed up again, so I didn't scare him off, which is great. But yeah, we obviously have to keep an eye on that.

Navigating Uncertainty in Clinical Research

It's an uncertain environment. I don't think you can pop open a news app and not see something about volatility, funding cuts. I personally, Jay Ferro's opinion, want to keep the United States on the forefront of medical research, but there's so much good research going on all around the world.

You just don't want to see any let-up in clinical research going because people are depending on it. There are life-saving, potentially life-altering cures or treatments in the pipeline that we want to prove that are safe if we can and then get them out into the people.

We're keeping an eye on it. Clearly, Chris, we don't want to see that type of impact. The clinical trial space, there's a long tail in it generally. These things are not done overnight. And hopefully I think we can withstand the chaos that's happening right now, and we get a little bit more clarity and predictability in the coming weeks and months so that we can get back to business. But right now we're keeping on keeping on.

Balancing Priorities and Saying 'No' Strategically

Michael Krigsman: Now is the time to subscribe to the CXOTalk newsletter. Go to cxotalk.com, join our community, participate every week. And we have another question, this time from Arsalan Khan, who is also a very long-time listener.

Jay Ferro: He is a regular.

Michael Krigsman: He's another regular, and we're grateful for Chris and for Arsalan and the other folks who listen and who just ask such excellent questions. And Arsalan says this: "How important is it to say yes, but also say no to all of your teams? You're juggling multiple teams and how do you decide when to say yes and when to say no and when to prioritize one versus the other?"

Jay Ferro: It is something that every CXO or head of IT or anybody that is responsible for software... Anybody in our space has to deal with. There is always more demand than supply. Always. I don't care what company. I have been a CXO now since 2008, and I cannot remember one time where I ever said, "Yeah, I'm good. I got plenty."

Ruthless prioritization is absolutely key, and you don't want to do that on an island. For me, what does that mean? It means partnership with my GMs who run our business units day-to-day. It means partnership with my CEO, naturally. It means partnership with my CFO, and he and I are absolutely joined at the hip with what we want to see.

We want to see near-term value. Certainly there are strategic investments. We're completely aligned on payback and making the decisions that have the most benefit for the company but also our customers.

Now, there are certain things that you just have to pay the dues every year. Cyber, you never want to take your foot off the gas with cyber. I am in an industry of trust. All of us, I think, to some degree are in an industry of trust. When that trust is broken, it takes a while to rebuild from that. Cyber is always pinned to the top of our priority list. Regulatory requirements, I think those always get pinned to the top. And then there are customer requests which come in and have to be juggled with those things.

But Arsalan, bottom line is, it comes down to ruthless prioritization. And I try not to say no. It's more of a not yet, or no, comma, here's what I can do. Can we agree on this in 2025, and in '26 we'll do a phase two? I really do my best to try to deliver some value even if it's a no. I prefer to be like a no, comma. No, but we have this. This gets you 40% of the way there. Can we start there and then put a plan together to pick up the rest of it in '26?

Michael Krigsman: It's not just a knee-jerk reaction, philosophical, "Well, this is something we don't do and therefore the answer is no." There's a reason. There's a...

Jay Ferro: Yeah, I will say this. If it's a customer request that comes in and it's just something that's not in our wheelhouse or something that's not a core competency or... Yeah, yeah. There are going to be times where you say, look, it's just, you have to be mature and self-select out of it. Or if it's a capability that just doesn't make sense for us to build, we are always going to try to find a way to partner or build or serve our customers.

If it's internal, nobody wants the house of no. I mean, Michael, you've been at this for a minute. You and I probably sat here 15 years ago and talked about the house of no and the CIO being the see-I-know and all of this other kind of thing. Nobody wants that, and I try never to say that. But it's always... When I do have to decline or just say, "We just cannot get to that this year or this cycle," it's usually followed by a comma, "But here's what we can do." And let's figure out a way together, GM or CXO, how we can tell the story so maybe we can get some additional funding to do this.

To me, it's about relationships and building and being transparent about how decisions are being made in the company and why we're choosing to do what we're doing.

Michael Krigsman: Okay, let's go to some additional questions. Joseph Pugliese on Twitter.

Jay Ferro: Ah, Joe up in Philly. Philly Joe.

Michael Krigsman: Okay, so friends. Great.

Jay Ferro: I know Joe.

AI in Healthcare and Clinical Trials

Michael Krigsman: And Joe says, "Will AI or other innovations lower the cost of healthcare? This is so needed in today's economic environment."

Jay Ferro: I think it has the potential to perhaps do that, certainly. We don't look at it that way. We look at it from a quality, efficiency, and privacy point of view. I think you're foolish not to look at AI and say there's not going to be some sort of cost arbitrage. We hear, we read about it every single day.

I think Bill Gates said something the other day, Michael, that he thought it would replace physicians or replace many doctors or healthcare roles. I don't know how comfortable we are with that just yet. I think we're at the beginning of our AI journey. To me, we still want humans in the loop. Given the regulatory hurdles and the importance of trust in the drug development process, humans are still very, very much in the loop.

And where AI plays a huge role is tackling things that maybe machines are better at: looking at large volumes of data, finding anomalies, assisting with document interpretation or image interpretation, excuse me, being able to make predictions, those types of things. But we're still at a point where I want a physician or I want an expert overseeing that work.

But yeah, I mean, the short answer to Joe's question is, I mean, if... there's got to be a cosplay down the road. I mean, AI isn't free, and... But to me, it's less about removing people. That's not the point to me. It's about getting our scientists and our amazing people at Clario to focus on a higher level of challenge and augment them with the technology.

Michael Krigsman: On LinkedIn, Greg Walters says, "When implementing AI in trials, is each application of AI unique to that specific trial or part of a standard AI process?" And he also asks, "Is the AI homegrown or partner?"

Jay Ferro: The answer is yes. The AIs are models that are... well, it depends. They are limited in functionality to a certain function, and that model can then be used across trials. They are trained. More often than not, they're closed models, so they are not continually learning, so they are closed as of a certain point. Whether that's a data privacy model that anonymizes medical images, it's a QC or quality control assist model that looks at early identification of quality concerns to help us ensure timelines, maybe a read assist where you're having a radiologist look at images, an MRI, a CT scan, those types of things, and it's identifying different things for you, and it can reduce inter and intra observer variability.

Those models are built and are fit for purpose for that function, but they're utilized across studies. And there's many more. We, at this point, we have 50 models, 50-plus models, that are proprietary, homegrown, across ECOA, cardiac imaging, respiratory, and precision motion.

We do have a proprietary gen AI platform that we have built internally on enterprise class... You can think OpenAI, Gemini, et cetera, that are private. We named her Claire. We're not super creative, Michael, so Clario to Claire. I know that's a... you can see the leap we took there.

Michael Krigsman: Well, hey, you're in a conservative business, the healthcare.

Jay Ferro: And if you saw the AI image, and I'll share it with you sometime, if you saw the AI image that AI created, we said, "What would Claire look like?" It is absolutely terrifying. It is terrifying.

But what I love about it is, a year ago, within the organization, if you had asked, I don't know, 1,000 people, "What is Claire?" They wouldn't have known, maybe a year and a half ago. Today, you're now hearing people say, "Well, we're going to put that in Claire," or, "We're Claring it." And it's become part of our lexicon because we've built capabilities to do a protocol summary, a user story writer, RFP and contract analysis, translations, document chat where you have a large volume of documents and you're trying to ascertain what exactly is... and so many more.

And the line of people behind our AI team with just use cases for Claire and what we're doing is huge. But going back to the question, we try to leverage third party wherever we can as long as it's safe, it's private. But a lot of our other proprietary models are built on our homegrown Aquarius engine.

Michael Krigsman: When it comes to gen AI and summaries, how do you ensure that there are no hallucinations? For example, summarizing a patient record, the AI could potentially think to itself, think whatever that actually means to the AI, but the AI says, "Given all of these symptoms, the patient must also have X, Y, or Z condition," and then inserts it. How do you deal with that?

Ensuring Responsible AI Development

Jay Ferro: Constant oversight, human oversight. First of all, we don't deal with anything that can then be attributed back to a Michael Krigsman or a Jay Ferro or anything like that. We deal with pseudonymized and anonymized data. First and foremost, we're not dealing with Joe Smith or Jason James or anything like that. We're dealing with pseudonymized and derived data, which is super important.

Number two, constant refinement of the model. We have an internal AI team under my chief AI officer who reports to me that I hired and brought onto the organization last year, Marko Topalovic. We are ruthlessly challenging our models to look for bias and hallucinations.

One of the ways that I think I'm really proud of is right out of the gate when I joined the organization and I realized that AI was a big part of our story, and we got a little bit of a running start. When I think when I joined, we had maybe one model in production that was scientifically validated, and today we're at 50 and growing, which is super exciting.

But one of the ways you do that is transparency within the organization. You want organizational literacy on what AI is, what it isn't, what we're doing, why we're doing it. More importantly, what we're not doing.

And we created a group right out of the gate, including my general counsel, Lauren Mishtal, my chief medical officer, Todd Rudow, my CISO, Murtaza Nisar. My chief AI officer, Marko, that I talked about. My head of quality, Todd, and all unified with our passion for the technology, but all coming at it from a different point of view and making sure that everything we're doing is built on trust, transparency, responsibility, bias control, et cetera.

And I urge anybody out there that is just at the beginning of their AI program, do not approach it as a CIO only or a CTO only. Don't do it. Okay? Don't throw all your stuff in an LLM and be like, "We're on AI, man. We're great." When cloud came out, you just started throwing stuff in the cloud and you wondered why proprietary information was suddenly gone. You don't want to do that.

Do it in a controlled way. And people immediately hear that, Michael, and what do they say? "Oh, my God. I involve attorneys and, whoo, that's just going to slow down innovation." Absolutely not. If anything, it has sped it up because there's no daylight between us and because our customers know that we're approaching it from a responsible point of view.

And the last thing I'm going to do is put a model into production that my general counsel, my head of privacy, my head of quality, my chief medical officer, my CISO, my CEO for that matter, of course, that we all haven't seen and are signed off on that we did it the right way and that we did it responsibly. If you're tackling this as an IT problem, I just encourage listeners to broaden, open the aperture a little bit.

AI in Clinical Trials

Michael Krigsman: We have another interesting question, again from Arsalan Khan on Twitter. How do you use AI on clinical trials and, of course, what's the impact? But he's also wondering, would removing AI guardrails and regulations make clinical trials better or faster?

Jay Ferro: You remove regulation, would they make them faster? Maybe. Would they make them better? No. I would argue that the FDA, and I have had numerous conversations as recently as this week with our colleagues over at the FDA, they are very bullish on the technology. They're doing it the right way. I think they're moving at a speed... I think people hear FDA and what do they think? Red tape. Grind to a halt. See you in three years. And that is just not true.

I might have said that coming into the organization five years ago. It would have been an ignorant judgment and I have learned that our colleagues in the FDA absolutely want to do what is best for people. They want to protect people, but they want these drugs and these therapies to be trusted, safe, effective, and they want them to improve lives. That's what they want. And in so much as AI can impact that, they want to lean into AI.

Now, the first part of this question, which I think is really, really good, is where can AI, where does AI fit into the clinical trial? Now, I'll give some examples where I'm seeing really good effectiveness in the trial process. We don't play in all of these roles, and I've talked a little bit about how we use it, but patient recruitment and retention.

Finding and enrolling the right patients in a trial is not easy. There's not a line of people hands standing, waving, hey. It can be time-consuming, expensive, difficult. Patient retention. You're taking a treatment or a pill or an injection or whatever over time that may or may not have side effects, that you may now have just real world things that impact your ability to participate in that trial.

And retaining patients is key and it's an important challenge. Minimizing dropouts is absolutely key. What can AI do about that? Well, first of all, it can manage or analyze large data sets, potentially electronic health records, genetics, social determinants. It can identify potential candidates who are likely to meet trial criteria faster, more effectively.

Using predictive capabilities, it could predict dropout risks so that those can be mitigated ahead of time where you can say, "Hey, look, there's a chance that Michael may or may not make the cut because we're seeing those behaviors." And that way I can intervene potentially and keep Michael in the trial. That's a huge application and ripe for AI disruption.

I think trial design and optimization, where we do play a role. Trials are complex, intentionally so because they're important and there's a lot of science behind them, and you want scientific rigor. You want meaningful results, but a protocol can be very, very complex. You're balancing patient safety, statistics, cost effectiveness, so many other things.

And AI can simulate different trial designs and predict outcomes with various configurations. Now, you wouldn't want to rely on that alone, but it can at least directionally move you in a path to optimize a protocol in looking for the most effective study designs. Not just cost effective, overall effectiveness, which ultimately would have the benefit of speeding up the development process, which is exciting. And those are just a couple of top of mind examples where I see AI making a huge difference.

Michael Krigsman: We have another related question that's come in from LinkedIn, from Lory Nouragov. Before we get to Lory's question, I just want to remind everybody that we have shows like this every week. Subscribe to the CXOTalk newsletter. Just go to cxotalk.com so you can sign up. We want you as part of our community so that you can participate and it's fun, and it's great. Sign up for our newsletter. And here is Lory's question.

Navigating Regulatory Challenges and Organizational Readiness

He says, "Does regulatory compliance create limitations for some of the more advanced AI use cases? If yes, what would those be and how do they work with regulators for making innovation more accessible?"

Jay Ferro: The regulatory environment being what it is, and keep in mind this is not the FDA only. You have the EMA and you have other regulatory bodies all around the world, all with their own criteria. The FDA is very thoughtful and very measured in what they do, as are many other regulatory bodies, filled with smart people who want to do the best for the most.

This is, to me, about partnership with the regulatory bodies. And I don't mean like co-development or anything like that. But understanding what the regulations are, understanding the hurdles and the barriers that we can overcome, showing them... I'll be honest. And the question is such a good one, and I think... And I don't want to come off as an FDA homer or anything like that, but as I've gotten to know them, I will say they want to do the right thing, and I don't see them at all as a, quote, "obstacle" or anything.

I think a bigger obstacle is organizational readiness. If you said, "Jay, is it the FDA or a regulatory body..." And to be sure, we are in a really highly regulated, and for good reason, we're in a highly regulated industry, and we have to be. Patient safety is absolutely paramount, and trust is absolutely paramount. And the science behind everything we do is absolutely important.

The reason I say that it's organizational readiness is because you're dealing with very complex... You're dealing with an industry that does not pivot overnight. We've seen it with healthcare over time. We saw it with fintech in the '90s and the 2000s, et cetera. If you had gone back in time and asked somebody in the '90s, "Hey, how safe is it to use a cellphone for millions of dollars of financial transactions?" they would've looked at you like you sprouted intel. "That'll never happen. You got to have people, you got to have whatever." And here we are today, moving money all around the world at a moment's notice.

It's going to change, but you're still seeing organizational resistance. I think in some ways for good reason, because there's a bar of rigor that you want to hit. Part of it is just culture change and making sure that people understand the power of AI in this truly unique fundamental force that we have. A lot of it comes from education as well, and overcoming some of the organizational inertia in older thinking.

I mean, look, I'm not going to be too hard on people, Michael, but I mean, if you had walked into a... and were hanging out with a 25-year veteran in the fintech space in 1994 and said, "I'm going to paint a picture of what the future's going to look like," they would've been like, "No. Let me tell you the 52 reasons why that's not going to work."

Michael Krigsman: Lory clarifies his question to say, how do you work with regulators to make innovation more accessible?

Jay Ferro: I was at a conference not too long ago. Lory, you'll appreciate this. And it was an FDA panel. And one of the FDA guys that... And I won't say his name 'cause it's not important, but he came off and I greeted him 'cause I just was blown away by the discussion, and it was all about AI and what they thought of it, and it just went in a totally different direction. And I said, "How do..." And I almost asked Lory's question verbatim, and I was like, "How do we partner with you?" "You can email me." And I was like, "Oh."

Proactive Collaboration and Regulatory Compliance in AI Development

"Oh, well, that's... I mean, that's interesting." We want to hear, we want to learn. But to put a finer point on it, I would say proactive collaboration, engage early. How do you do that? Well, talk to them. Okay, this isn't an ivory tower that... I will say every FDA person or every regulatory person that I've ever talked to has been willing to have a conversation. Establishing clear communication and being proactive with the regulatory authority, whether it's the FDA and EMA, et cetera.

Number two, understand regulatory frameworks. Show that you're compliant with existing guidelines, making sure that you're familiar and you understand what they're trying to accomplish, and that you're not coming at it from a purely tech point of view and be like, "Yeah, but if we could just release the guardrails and let this thing go nuts, imagine all the good." Probably some truth to that, but I think the reality is that the regulatory frameworks exist for a reason. They change over time. Guidance has changed over time.

If you had said 20 years ago, "We're going to have an AI reasonable use policy," I think most companies would've been like, "What? Why the hell do I need an AI reasonable use policy?" Making sure you're familiar with existing guidelines and that you're adapting your AI to fit that compliance. I think that starts you off in a very good conversation with regulators.

Making sure that you emphasize safety and risk management with regulators. The FDA often requires risk management strategies. They don't want to hear about the art of the possible without the other side of the coin. What are you doing to manage risk to ensure that the AI technologies that you're proposing are safe and that any potential risks to perhaps the patient, the study, et cetera, are mitigated?

And the last thing I'll say is develop a regulatory-friendly development practice, which I harped on at the beginning. Following best practices, having SOPs in place that can pass regulatory scrutiny, transparent in training models, bias mitigation, all of those types of things build trust with regulators.

Transparency and Cross-Industry Collaboration in AI

Michael Krigsman: And on this topic of transparency, Arsalan Khan from Twitter comes back and he says, "If we want transparency across industries, should companies share their next AI with others so collectively whole industries can get better? And what about the role of gatekeepers in all of this?"

Jay Ferro: I do think there's an opportunity across industries to promote transparency and drive positive outcomes across the industry, whether it's healthcare, life sciences, fintech, et cetera. I think there are some benefits of doing that. Certainly speed, collaboration for innovation, and you're going to get more faster generally.

Public confidence, I think, when they see collaboration and trust and transparency across industry. There's certainly an opportunity for better regulation, and you're seeing that already with Europe. Their AI regulations are not just for life sciences or healthcare. It's AI in general. Matter of time before the United States does it, either at the state level or the federal government level. Certainly, you have some draft guidance already out there. But I think it speeds that up, and I think you get a higher quality of regulation when there's more collaboration and sharing. There are a lot of reasons to do it.

Michael Krigsman: I'll just mention for folks that are interested in Europe, European adoption of AI and regulatory efforts, just a few weeks ago, we had two members of the House of Lords discussing these exact issues here on CXOTalk. These are folks who are creating policy in the UK. If you're interested in this topic in Europe and the UK specifically, just go back and look at CXOTalk. We had... It was a really great discussion.

Jay Ferro: Well, I'm sure... I mean, look, they are always at the forefront of regulation. And I don't mean that in a snarky way, but I mean, I think they've done a nice job in balancing innovation with risk mitigation, transparency, and accountability. It's not perfect. Nothing ever is. But the EU AI Act... I mean, they came out of the... I mean, what is that? In '21, I think they launched that. And which is crazy. It was very, very well thought out and what I like about it is it triages risk on acceptable, high, limited, minimal, et cetera. This risk approach where, hey, if it's a system that poses little to no risk, it's an AI in a video game, it's an AI in a spam filter, have at it. You're going to have to check a few boxes.

Industry Similarities and the Role of Technology

You're going to have to do a few things. But yeah, we don't want to slow stuff like that down. And it seems to me they at least try to right size the regulation, which is good.

Michael Krigsman: We have another question from Greg Walters, who is aware that you have a background in cement. You can tell us about that. And Greg Walters wants you to compare AI to, quote, "cementitious materials," given the fact that concrete has been around thousands of years?

Jay Ferro: Outing me like that, Greg. My word.

Every industry is more alike than not. I know we like to not think that. I think every industry thinks it's a perfect little snowflake, Michael, that is a unicorn, every little thing. And to be sure, every industry has nuance and has its own lingo. It's got its own ecosystem, and it's got its own special spin on regulation or whatever.

But I can tell you, the discipline of product management, technology management, financial management, software development, security, et cetera is 80% the same across industries. If you look at manufacturing, my time as the CIO of Quikrete Companies, and even before that, at AIG, or the American Cancer Society or EarthLink, you were doing many of the same things.

Now, the size, the scale, the severity may be different, but when are you not protecting customer data? When are you not trying to deliver high-quality results on time? When are you not trying to leverage emerging technologies to deliver a higher quality product, or have an amazing customer experience, or improve operations or drive revenue?

And I would challenge CIOs and CTOs and aspiring CXOs to think about things that way and not box yourself into one particular industry. But how is cement like AI? I know there's a punchline, Greg, somewhere out there, and I don't know what it is. Maybe somebody will suggest it, but...

Purpose-Driven Technology and AI in Healthcare

Michael Krigsman: Joe P. on Twitter comes back, and he says, "How do you keep your team focused on the outcomes and avoid becoming enamored with the technology? In other words, avoiding shiny new object syndrome."

Jay Ferro: Education, transparency, building financial literacy, education, educating the team on why we exist as a company and what that really means? Those are all a lot of really flowery words. Site visits, discussing with patients, showing them what we do, visits from different departments, building those relationships so that people, no matter where they are in the organization, in my team... I don't care if you're a health test level one, I don't care if you're a coder and you're a junior person that just joined the team, I want them feeling the importance of what we do.

I want them recognizing that there's a patient at the other end that is using an app on either on a provision device or a phone or is using a connected device, whether that's a blood pressure device or a respiratory device in-clinic, that we built or that we built an interface for or that we have AI integrated with. I want them understanding that what they're doing has real human impact around the world.

And when you do that, yeah, you're still going to get excited about the technology. It's what we do, but it's technology with purpose. It's not technology for technology. We try to do a lot of education, Michael and Joe P., but keep people focused.

What does that mean in practical terms? Site visits, site education, voice of the customer type activities, making sure we're hearing from patients, making sure we're showing teams the impact of the technology that they're building and that they're delivering, whether that means hearing directly from customers, et cetera, so that they understand the impact. It wasn't just a bunch of code on time. Which is nice, I want that, but there's purpose behind it.

Michael Krigsman: On this topic, Arsalan comes back again, and he's wondering about the future of AI, let's keep it focused on clinical trials, and he says, as AI becomes increasingly more of a commodity.

Jay Ferro: If DeepSeek has showed us anything, it's that I think the idea of a trillion dollar, one size fits all, one magical product to rule them all is not going to be long for the world. I think there will be some commoditization of technology over time, and I think what you'll see is far less one giant model and far more fit for purpose.

What... The future I see, I see some amazing things happening in the future, particularly in healthcare, in clinical trials, keeping at speed, improving in quality, patient experience. What does that mean? Well, it means if a patient is having a bad day or there's just... There's just some empathy that even models today can show. More time with an investigator or a physician if it's healthcare.

Physicians will tell you constantly and nurses will tell you constantly, I'm sure everybody knows this, they spend more time entering things into an EMR than they do with a patient. EMRs are designed for one thing. Or two things, to bill and to keep street legal. And I'm not trying to be depressing or snarky or anything like that, and I don't mean anything bad toward our friends at the big EMR companies, but there's a huge opportunity for AI, whether to listen to the physician as he's talking, to transcribe notes, where they can turn around and spend even another minute or two with a patient and improve that patient experience. I am really excited about what the future holds, particularly for patients and better outcomes.

Michael Krigsman: The through line of our conversation seems to be customer experience and broaden it, patient experience, whatever domain that might be.

Jay Ferro: I think that's true. Look, if not them, then who? I mean, if you're not doing it for them, then why are you doing it? And look, you and I, if I take off the motherhood and apple pie hat and say, "Okay, Jay, yeah, patient experience is super important, but really, companies exist for shareholder value and to make money," fine, you're not wrong.

I would argue that one of the best ways to do that is an amazing patient experience and an amazing customer experience. If I'm making my sites happy, or at least less mad at me, or am I providing a frictionless experience, and they always know at the other end there's a company or a team that is working hard to do the best. We're not always going to be right. We're not always going to get it right. But we're always doing what's important and what's the best for them and for their patients.

If they know that, I would argue that they're going to want to continue to do business with Clario, and they're going to continue, want to continue to partner with us. I think the two are inextricably linked. I think you can have an amazing patient experience, an amazing site experience, and sponsor experience maintaining scientific rigor, high quality, privacy, speed, and all the other things that we want to do, and you can still make money as an organization and build shareholder value and do all of those things. And I would argue that if I can do the first one really, really well and I'm making good choices, the second one will come.

Scaling AI Projects and Overcoming Adoption Challenges

Michael Krigsman: What advice do you have to business and technology leaders in the enterprise when it comes to developing AI projects and scaling those projects as you have done?

Jay Ferro: Don't do it in a silo or, as my old boss used to tell me, "A highly polished cylinder of excellence." Look, partner. Talk to your partner ecosystem, other CIOs, other CXOs. Talk to your suppliers. Understand, educate yourself. Talk to your peers in the organization and work with them, not against them. Whatever you do, don't make it a tech-only endeavor. In no way does that minimize the importance or the excitement about the technology, but I promise you, if you can do all of those things, it will speed up what you're trying to accomplish. And as always, Michael, and I think I said this on our first interview, think big, start small, scale fast.

Michael Krigsman: Think big, start small, scale fast. Great advice. Hey, we have one last question that's come in very quickly from, again, from Lory Nouragov who says, "Do you feel like we have the right people, training, mindset, the site, patients, nurses, et cetera, to enable and provide all of the data needed for the AI applications? Or is the technology far in front of actual adoption?"

Jay Ferro: There's a huge educational hurdle that we have to hit. There are people at the sites and with the patients. I think they're just handed a lot. They are handed a lot of different devices. And it... And I promise that it's not because they are not smart people or hardworking people. It's just a lot.

And if... I mean, the first part of the question is, do we have the right people? Yeah. I mean, the ecosystem of investigators and sites are amazing. And they work their tails off, and they're more than capable of picking this up. I think it's on us to train. It's on us to show them the power. It's on us to make it simple.

I think it's going to require patience, Michael. I do not think we're going to be able to hand them a magical calculator and say... And have them be enamored with the tech in one day. It's going to take some time. And we have to be, I think, a little patient as we change the paradigm. But I'm optimistic that we have the right people. We just got to meet them where they are.

Michael Krigsman: And on that note, Jay, a huge thank you. I'm so grateful for your taking the time to be here with us today. Really, really appreciate it.

Jay Ferro: My pleasure, Michael. Thank you for having me.

Michael Krigsman: And thank you to everybody who watched. You guys who ask such amazing questions, you are so smart. Now is the time to subscribe to the CXOTalk newsletter. Go to cxotalk.com, join our community, participate every week, and we'll see you again next time. Take care, everybody. Have a good day.