GlobalEdgeTalk

AI Readiness In Healthcare, Without The Hype

• Alex Romanovich

Share episode

Send us a text

The pressure to “do AI” in healthcare is real, but shipping a model isn’t the same as changing care. With Dr. Saima Anis—physician turned public health leader and enterprise IT strategist—we unpack how hospitals can adopt agentic AI responsibly while actually making clinicians’ jobs easier. We start with the human questions a good diagnostician asks: where is the pain, who does it hurt, and what outcome matters? From there, we build the case for culture, literacy, and trust as the groundwork that makes any system stick.

We break down governance without the jargon: data lineage and quality, explainable models, stage-by-stage audits, and human override by design. Dr. Anis clarifies why agentic AI is not a chatbot or an RPA script; it’s a domain-trained, continuously learning framework that can draft notes, synthesize evidence, and propose structured differentials—if and only if it is integrated into real workflows with clear guardrails. We explore how to reduce cognitive burden for physicians, where to start with repetitive tasks, and how to prevent drift and hallucinations when agents collaborate.

Leaders will hear specific pitfalls from 2025 to avoid—vendor lock-in, FOMO-driven deployments, and loyalty to failing pilots—and practical habits to adopt in 2026: stakeholder alignment before tooling, measurable outcomes, and transparent audit trails. We also talk candidly about leading as a woman in a male-dominated field, sustaining momentum under pressure, and why innovation is not synonymous with AI but with creative problem solving that delivers value. If you’re planning a rollout or rescuing one, this conversation offers a pragmatic playbook to move from hype to helpful.

Enjoyed the conversation? Follow, share with a colleague who’s wrestling with AI strategy, and leave a review with your top takeaway.

Support the show

Alex Romanovich:

Hi, this is Alex Romanovich, and welcome to Global Edge Talk. Today is December 11th, 2025, and we're joined by Dr. Saima Anis. Hi, Doctor.

Dr. Saima Anis:

Hello. How are you, Alex?

Alex Romanovich:

I'm doing great. Thank you so much. You have two decades, almost two decades, or maybe even more, of experience across healthcare systems, public sector initiatives, advanced AR architectures, probably not 20 years, but uh, you know, some of that experience. And you became kind of a trusted authority with a number of organizations that are transforming uh complexity into clarity and reality within the healthcare and life science industry segments. You also are managing director and partner of your own company, your own consultancy called Ogmind.ai. And you were previously with DV8 Systems, Info Systems, working globally. Tell us a little bit more about your background. Maybe I'm not covering it completely, but what led you to this intersection of informatics, data, medicine, healthcare, and now AI?

Dr. Saima Anis:

So thank you for having me on this podcast. I feel honored that you want to highlight my career path and what I've been up to and what I'm doing currently. My journey has always been mission-driven. Starting with my clinical training in Pakistan, I was um really exposed to a lot of stark um inequities and resources and the gaps. And that made me realize that I wanted to address the systemic issues in healthcare. And really, you know, there was not one moment where I would have said that, you know, I I could do this right, but it was a growing realization as I went forth uh with that training. So, and I was thinking that perhaps um having a greater impact meant that I pursued something other than treating patients one-on-one. So that led me to a public health scholarship in the US over 25 years ago. And I've just been learning and evolving since.

Alex Romanovich:

And it's not just some of the research work and informatics work you've done with um as of recent AI. You've been involved with a number of hospital-based initiatives, specifically with the Jefferson University Hospital System. And then you led a major rollout, mobility and unified communications rollout across 14 different hospital systems. Tell us more about that.

Dr. Saima Anis:

Yeah, so my reinvention started a long time ago, like I said. So I decided to move from clinical to public health, and it was very deliberate as like I wanted to step outside the realm of traditional medicine and into a world of systems and data. So I already became known as an expert in public and population health in the city of Philadelphia from the health commissioner's office. So the single reinvention came to a conclusion with IT and health systems with a large hospital system on the East Coast, 18 hospitals at the time. Now it's much more so. Um and it was driven to solve those complex problems in IT, and I started gaining multiple perspectives from informatics and from the AI architecture, and that's where I am now because of all of that.

Alex Romanovich:

You know, um I attended a healthcare conference recently in New York City, and I was talking um offsite with um one of the um chief technology chief information officers in the major hospital system, about four or five hospitals. And we talked about AI, and I asked him a question. I said, you know, how are you adapting AI and rolling out AI within the hospital system? And he said, you know, we're not even approaching it yet, because if I did that, if I said that, we would have a lot of fear, a lot of anxiety, and frankly, we may not be ready. How do you think, what do you think the leaders should be ready for? And what new risks should the leaders within the hospital system, within medical field, within healthcare, what should they be paying attention to?

Dr. Saima Anis:

I'm glad that you asked that because one, AI is an elusive word right now for a lot of organizations, especially those large, multi-complex systems, right? And although it's treated as technology, it's really just not about technology. It's there's a huge, big challenge in it that is the human element. And the technology is often the easy part. So the hard part in putting people together and what overwhelms these systems, and we're not, like you said, they're not even touching it, is aligning those stakeholders and changing those long ingrained like workflows and uh building trust, you know. And a lot of people have different levels of learning and they have competing priorities. And so getting everybody on the same page for the literacy, for the cultural resistance, those are the kinds of things that are needed to be done before anything else, just to add to your comment about they're not even thinking about it yet, right?

Alex Romanovich:

So it's about Well, yeah, and and AI is gonna be all about speed and quick adaptation and automation and so forth. But yet a lot of the organizations are kind of pausing and saying, oh my goodness, you know, let the let us make sure that, first of all, it's safe. We can trust it. Number one, and number two, what is this going to do to our jobs? What is this gonna do to our organization, our culture, and so forth?

Dr. Saima Anis:

Yeah, it's like uh I think that uh alignment of the people in the organization will beat the brilliance that AI can bring to any solution anywhere. Healthcare is just one domain. AI is like almost domain agnostic because it can work itself to train itself in different areas, right? So you can have the most elegant technical solution, but you have to learn to be a translator along the journey and get everybody on board to do that. Empower those champions within those groups, right? So it's not a magic bullet and it can't fix all our problems until you know how to integrate it well within the culture, within the system.

Alex Romanovich:

So how do you install the governance over these systems or maybe smaller projects? And then how do you scale it across the entire enterprise? Well, is there is there a, you know, is there a methodology that that they can follow?

Dr. Saima Anis:

Gosh. Governance is another elusive concept for a lot of organizations. You know, they think that having your um processes and boards in place and having the principles in place will ultimately trickle down to the governance effect, but that's not what happens. Um, you know, you it's not about this like you use the word speed, right? It's not about moving fast and breaking things. It's not about just doing it. It's about doing things intelligently, iterating the intelligence framework. For instance, if you've got to work to build something and pilot something, at each segment of that building process, you have to have the governance in place. And by that, what I mean is you have to have those mechanisms that control, audit, and control those processes. And very rarely do people realize that at each path of AI implementation do you have to do that? Not just a top, it's not just a top-down approach. The second thing is the ability to track the risk assessment. Are there risks and biases before a project is even started, right? Do we have the data, the truth of the data as a truth, you know? Where is it coming from? What's its lineage? Are we able to track it back to how it's going to be used and now where it's coming from? Are you able to clean it up? You know? And then of course, can you explain the models based on that that you're building for? So governance oversees a lot of all of these things without people realizing that at these institutions. And the message then is that should you just go in or should you go in and implement these, or should you have an overarching human in the loop as well? And somebody who can intervene and override the AI when necessary, that all needs to be put in place. And that's what governance is about.

Alex Romanovich:

When you consult an organization, medium-sized or large organization, and the organization might be considering doing some AI projects, or maybe they're already doing it. What are some of the first diagnostic, you know, three to five questions that you would ask them in terms of what have achieved, what they have achieved, or what they're about to achieve, how they structure their pilot systems, how they're planning to roll them out. In other words, how do you diagnose quickly whether they're going to be successful or not?

Dr. Saima Anis:

That's a great question, Alex. You know, for diagnosis, if I were to go back as a clinical person, I would think clinically and say, well, tell me how you feel. How how how's every day going for you? Where's the pain point, right? Similarly, what is your problem? Explain the problem in detail. Where does it hurt you? What's your pain point? And secondly, very importantly, it's as important as the first question, where does it pain you? Who does it hurt? Who in that whole entire process of the business and the human context are we trying to solve for, right? So those are the two big questions that I ask, you know. And of course, you can't detach them from who else would be a specific set of like people who will gain an advantage from using the system. So Dan in itself is a very important question. So the human element again, because you're building to ease for people, you're not building to make it easy for machines only. The end user is the human consumer. And so we're almost like going with the mindset but that we're trying to augment people's work. Because you also mentioned at another point that you know people are wary of losing their jobs because of AI. If we use and we consult with AI as a tool, a fruit that will help us and augment our work rather than replace our work, then we're more comfortable. And so the trust factor also builds in right there. So those are the kinds of approaches you deploy when I go in as a consultant.

Alex Romanovich:

I love the analogy of you asking it as a physician versus asking the organization uh uh some very specific questions as a you know technology and AI diagnostician. It's actually a great parallel. Now, we've all heard about agentic AI. It's a probably most commonly used term in any industry for that matter. Hospital systems, large organizations, healthcare organizations, excuse me, and even some of the research organizations, let's say in biotech or life sciences or what have you, uh, you know, pharmaceutical even, they're beginning to use agentic AI. They're beginning to use agents to begin it to deploy those agents. How do you think those agents are going to reshape the organizations in 2026 and move from experimental to production, if you will? And where do you think those agents are going to be mostly used? Um, you know, in terms of functionality, in terms of features and uh value, and also in terms of securing um and preserving maybe good data and uh preserving good business rules and good code, or maybe even transforming that code into even a better code. Tell us more.

Dr. Saima Anis:

So, yeah, so agentic AI is a term that's been freely used, and a lot of very clear misconception that people have that I will remove here is they think RPM processes that they can monitor and rapidly automate workflows, workflow automation is akin to agentic AI. It's not. Chatbots are not agentic AI. Agentic AI is a proprietary or non-proprietary framework that is deployed and trained on specific systems and deployed in the cloud or non-cloud premises, where it can be trained in a specific domain to do the specific task that it's trained to do. It's continuously adapting, it's continuously learning, it's continuously training itself based off of like very basic uh ontological texts and information. And then on top of that, it will learns from real-world evidence and the questions that you ask it or the tasks that you perfor ask it to perform. So going in with that clear conceptual understanding, it becomes a very powerful tool. And but it's only, like you said, data, right? It's only as good as the data that you train it on. And then again, after you train it on the data, it's only as good as the workflows that you integrate it into. So for instance, in healthcare, you want to clean up the task of physicians writing notes or coming to a conclusion about their diagnosis. If the agentic AI is trained on reading ontological text as physicians are have been reading in med school, for instance, right? You feed it all that literature, it'll absorb it, it'll eat it up, and then by your continuous interaction with it as a physician, it'll learn to adapt to the formula of how to come to a differential diagnosis. So that's one good use case for it, right? So it'll help get the cognitive burden and the cognitive load off of your mind and free you up as a physician to think clearly and reinforce your thinking methodology because it's reinforcing what you're thinking and it's giving you the right answer. All right, one, two, three, four, five. These are what could be wrong with the patient, right? So let's present that. In this case, it's painstakingly working one to sift through the data that you would be using in your mind to come to that conclusion. And it's also sifting through its own data mechanisms, right? So those pathways have to be very safe. They have to truly enhance the ability of these professionals that are using them. When you use agentic AI comes the whole process about the whole thinking about the framework is trust. Can you trust it? Can you trust it enough? If you use it enough and you train it enough, it almost becomes autonomous in its thinking. So that's where you have to put a guardrail. Again, human in the loop process to oversee guardrails in place to see that it doesn't exceed the thinking capacity or doesn't exceed the work that you actually asked it to do. Because at one, in one instance, agents also talk to each other. So you have more than one agent in the place, they learn to communicate, and they may start giving you an answer that is not close to the truth that you're seeking. It may, what the common terminologies start hallucinating. And so be wary of that, while at the same time, be very receptive of the fact that it's just something that'll help you reduce your cognitive burden. That's one way to view it.

Alex Romanovich:

Yeah, it's a great analogy and uh great use cases. It seems to me, however, that there are so many areas of mistakes that could be made by the organization. Um it's still a very fragile environment, very experimental environment. The CEOs, CIOs, CTOs of these organizations that are deploying it fairly quickly, what are some of the mistakes they have been making in 2025 that you're aware of? And what are some of the mistakes they could be avoiding in 2026?

Dr. Saima Anis:

Well, if you're talking about mistakes as um, you know, things that they go back and iterate on, it's to deliberately reflect on the lessons that they're learning from deployments or they're learning from adapting technology. And those could be at the end lay at the intersection of project management, they could lay at the intersection of highly technical domain implementation, or they could lay at the intersection of keeping centered. So for instance, like if I had to say I was with an executive board and one of my failures was that it was people perception, political and cultural alignment. Organizations such as, you know, larger organizations really have to learn this painful lesson that I I learned in my experience. And I don't know if I should have declined the project for that reason, but really you can't stay wedded to a project for the sense of loyalty if it's not producing an outcome. So organizations have to move past a failure, learn from it, and recognize that there's wisdom in knowing that the season's ended and move forth. That's one of the things that I would stress. The other thing is the con there's a constant need that people think they have to prove themselves. And they organizations think they have to prove themselves by adapting. And if they don't adopt a newer technology, because AI is literally seen as that, they haven't jumped the bandwagon, they're going to be left behind. It's like the FOMO, the fear of missing out. Whereas if you adapt technology to suit the cultural needs and the actual physical workflow needs and organizations, that suits their purpose better. So go slow, approach with caution, get everybody on board, stakeholder alignment, and then the technological concept and the technological piece comes right in. That's really important. So a conscious effort to model that and show that as a vulnerability. If you're stuck in something, then you ask your peers. High C-suite level executives will make a mistake of introducing vendors to their organizations and then be stuck with it because they don't want to claim claim or take responsibility for a certain product not working out. Something to watch out for.

Alex Romanovich:

Yeah, absolutely. Now you're you have so much knowledge, so much experience, and you're operating still in a very uh male-dominated world of IT. Because of your highly technical domain, your skills, have you experienced any challenges being a woman as an example?

Dr. Saima Anis:

I mean, so gender parity in technology is still something that they're striving for. And I think I I have an advantage which I didn't view as a strength before, but just the fact that I'm able to take the most strategic aspects of my skills and training from clinical to technology to AI to public and population health and see that as a strength makes me oversee the fact or overlook the fact that I might not be the best player. But yes, as women, you do have to prove yourself a little bit harder. It's like uh from um man, if their their leadership skills, they're assertive, right? And that's so they're seen as leaders. For women, they're seen as pushy. But really, it's if you if you kind of like overlook that aspect of it and you think about it in terms of what can I do to make myself better, and that's what's going to prove my worth, then sometimes you have to iterate with a vulnerability in mind, which is to say, well, if I'm stuck at something, I'm going to make it a point to ask my team members. And that also builds confidence. And so that's been one of my successes, the way I practice.

Alex Romanovich:

I was just gonna ask you, what's some what are some of the habits or philosophies that have helped you to be well balanced, centered under that level of pressure?

Dr. Saima Anis:

Yeah, so it's I've learned that it's not the smartest person, but it's the person that keeps the momentum going. And so if you stop, you lose it. If you don't do it, you lose it. If you do it, you keep going. And that's where my greatest strength lies.

Alex Romanovich:

Aaron Powell, but also being honest with yourself. So let me ask you this question. Um, if you met yourself uh 10-15 years ago over a cup of coffee, what would be your advice to Saima of that time?

Dr. Saima Anis:

Which it would not probably change from what it is right now. Don't let small setbacks phase you or fade your energy. Just keep take them as a stepping stone and a sounding board for future work. And every career faces challenges. Every field in technology and non-technology faces its own inertia and ch challenges. You just have to keep looking at the bigger picture as you go along. Also, Margaret Meatman said, never doubt that a small group of committed individuals can change the world. It's usually the it's the only thing that ever has, right? So come come get yourself together with like-minded, mission-driven people, and you're going to go a long way.

Alex Romanovich:

Networking is important. Great advice, great advice. Let's talk about innovation for a second. It almost seems like the um in the modern enterprise or even startups, because of AI, innovation is almost synonymous with AI and vice versa. I mean, that's what I mean, investors don't even want to talk to you as a company or a group of individuals, so you don't have AI spelled, you know, ingrained in the definition of your company, if you will, or your philosophy. Is it true that innovation is synonymous with AI? And is it is that a measuring stick, so to speak?

Dr. Saima Anis:

Innovation is more synonymous with the tools that you're using to deploy to make a project a success. If you want an investor to invest in your product or your company as a startup, you really have to look at it from what are you doing differently that nobody else has thought about. Innovation in that sense is what carries value over just offering the AI as a solution, unless the solution has an ar proven ROI, which as a startup you never really have at the start at the right. So you're going to take some time to prove it the ROI on that investment. But are you at least trying to do something that nobody else has ever approached it from those eyes? You know, from a very creative lens. So creativity and innovation, in my view, go together. And people get taken in by just the word AI. It's a buzzword, like you said.

Alex Romanovich:

Well, yeah, it's a buzzword that obviously means a lot of different things. And in smaller organizations, there are companies that are built on AI stack. They they start with AI, they they they're basically using artificial intelligence as the foundation for their firms, not just their products. Now, what about the large enterprises? There seems to be a notion that AI is going to replace people, especially if it's a publicly traded company. Wonderful, we now found a way on how to be even more cost-effective, more automated. That's all going to be great for the stock. It's also also going to be great for our shareholders. But are we forgetting about the human in this entire equation?

Dr. Saima Anis:

I think it's important to recognize that you should not forget about the culture, the process, and the people, very importantly. If you keep in mind that the people are the ones whose work is augmented and they're not replaced by the AI, that's what's going to make AI a success in the organization for the workflows that they're looking to replace. And you may never completely replace the human. It's also having an AI skill for a human worker in the organization, it becomes a great equalizer. You can be part of the team talking at the same level as the CEO or the CEO because he has access to the same AI that your other workers have. So you can actually have very productive one-on-one conversations and find solutions to the problems that ail you in that organization. So you can't underestimate its power, but you cannot also underestimate how humans are deploying that to make an effect.

Alex Romanovich:

Of course, there's some organizations like large banks and some of the large companies in regulated industries that internally almost forbid the use of AI in the same way that uh high schools are beginning to ban AI. I think it was recently mentioned that Australia is now banning social media for high schoolers and school students. It's been a fascinating conversation, Saima. What are your parting words? What should be the advice? What is your advice to some of the organizations, early stage companies that are thinking maybe even basing their strategies and the products on AI? And some of the more mature organizations that are exploring, trying, testing, what should be that advice?

Dr. Saima Anis:

I I think uniformly across the board, my advice to any sized organization or in any domain would be to see agentich systems or just AI as being tools that we use to uh bring to our team that we can collaborate with rather than replacements for people. And so in that sense, have them think about automating repetitive tasks that bore people. Something that can be mindlessly done, free up that headspace for your employee and give it to AI to do. But also, if there's complex workflows, you still need human decision making and human reasoning in that loop. So bring bring in your employees for that. The key is to design the systems when they go in to augment rather than replace human intelligence and make them free to exercise the creative aspects of their work. So, lesson learned, creativity goes a long way with using AI.

Alex Romanovich:

Excellent. Great. Thank you so much for being with us. I'm sure this is not the last time we're going to be talking and uh wishing you all the best. And I know that those organizations that you're involved with are in great hands. Thank you so much.

Dr. Saima Anis:

Thank you so much for your time, Alex.