00:01.86 Ratnadeep Bhattacharjee Hi everyone, and welcome to another episode of Leaders Perspective, the podcast where we bring together visionary voices at the intersection of healthcare, technology, and innovation. 00:13.29 Ratnadeep Bhattacharjee Today's conversation is one I have been particularly looking forward to. It's titled, Role of Agentic AI in Healthcare care and Life Sciences. And we'll be exploring the evolution of AI from not just task automation, but to intelligent orchestrations and whatnot, right? Especially in the highly specialized world of life sciences and clinical development. 00:33.64 Ratnadeep Bhattacharjee My guest today is none other than Dr. Lana Fung. I hope I'm pronouncing it correctly. A true pioneer in this space. 00:39.53 Lana Feng You're correct. 00:41.92 Ratnadeep Bhattacharjee Lana is the is as currently working on a platform that is redefining how domain experts interact with AI. not through lines of codes, but through intuitive, plain English collaboration with modular AI agents. 00:57.01 Ratnadeep Bhattacharjee This sounds interesting, i'm pretty sure for our viewers, right? She and her team are building a system that allows healthcare and life science professionals to describe what they want in and a natural language and have modular intelligent agents assemble ah systems behind the scenes, right? 01:13.68 Ratnadeep Bhattacharjee ah Dr. Fung brings 20 plus years of leadership across biopharma, genomics and health tech, including senior roles at Genentech, AstraZeneca, Novartis. 01:26.80 Ratnadeep Bhattacharjee She holds a PhD in molecular biology from the University of California and is deeply focused on accelerating insights to impact across the life sciences value chain. Lana, it's an honor to have you with us. 01:38.41 Ratnadeep Bhattacharjee Welcome to Leaders Perspective. 01:40.47 Lana Feng Oh, thank you so much. Thank you so much for the introduction. a little bit of a correction is that um I was actually ah not part of AstraZeneca and Roche and what have you. 01:51.81 Lana Feng They were our clients when when when I was the co-founder and CEO of Huma.ai. But my background did come from pharma and precision medicine. Like you said, I spent 20 years in that space. 02:04.11 Lana Feng um I was a head of ah biopharma at a oncology company, Comgenoptics. and um and build that from our very rapidly. And then leading to the acquisition by Novartis. That's how I ended up into global pharma. 02:21.13 Lana Feng So, um but I've as kind of head of human AI, I spent um a number of years kind of pioneering genitive AI, verticalizing it for healthcare life sciences, and um' collaborated with OpenAI eye in 2022. 02:36.12 Lana Feng We launched a platform as ah be before they actually launched ChatTBT. So lots of experience in leveraging AI for highly complex and highly regulated industry like healthcare life sciences. 02:49.28 Lana Feng So i'm pivoting into this new company called Coxia is really kind of can we take AI even further, right? um Because although I will kind of talk about it some more in terms of really leveraging agentic and then pushing it even further and using this newest um um ah ah Neu-Wing Li-Wing Li-Wing Li-Wing Li-W model context protocol and cp and then by kind of um putting together ncp components 03:20.13 Lana Feng allows anyone with an idea to be able to build ai native applications that is production great and deployable really kind of crossing this prototyping to production chasm 03:34.04 Ratnadeep Bhattacharjee Honestly, sounds very interesting, Dr. Lana. And, ah you know, let's let's go back a few steps right now. Let us start from the beginning, right? What really inspired you to pursue the vision you're building towards, right? You have spent years in both big pharma, you know, even early stage health tech. 03:53.94 Ratnadeep Bhattacharjee But what problem did you really see that you felt you had to solve this? 04:00.41 Lana Feng The problem we're trying to solve really had the very high level is says how do we better serve patients, right? And as we all know that the um our healthcare care system is the sick care system, right? 04:14.96 Lana Feng It's quite broken. It's um lots of friction and and very inefficient, right? And it's because it's industry is risk adverse and it's because we're we're you know we're talking about saving patient lives. 04:29.51 Lana Feng So that the the risk is quite high. So therefore we're very cautious, right? We're always kind of the last to adopt technology. Life science is a little better because it is kind of the innovation arm of the healthcare care system, right? You're thinking about providers, you're thinking about payers, and you' thinking about life sciences companies. 04:46.95 Lana Feng bringing innovative therapies or devices or treatments right too too medical to medical care to health care so that's really kind of think about this ecosystem and um the reason i'm i'm i'm doing this new company is that um we all have as because i'm not a coder right so i'm more of ah a a stakeholder subject matter experts in this kind of grand scheme of things and um you know, I have, I need to kind of do this in order to help and innovate in healthcare life sciences, right? i have ideas of how we can do this. How can we leverage identity AIs and what have you, but I'm not a software developer, right? So traditionally, what we have to do is ah in a company, particularly if you're you know small to mid-sized companies, right even global companies, global pharmas, they have all the resources at their fingertips are having challenges. right If you think about these global pharma companies, they have their IT department and they have all the business stakeholders within 05:49.91 Lana Feng kind of from drug discovery to clinical development, all the way to, you know, medical affairs and commercial and post-market, right? So that's kind of majority of the company and IT team typically is already overwhelmed, right? 06:03.86 Lana Feng And the limited resources, and they typically tackle the enterprise wise, like maybe top five initiatives. right And then there's these the SMEs in the business side really left kind of to to think how do we kind of leverage the resources we have right kind of outside IT t to be able to innovate. 06:24.71 Lana Feng So this is kind of um really what we're kind of targeting is that can we allow these people It could be HR, it could be finance, right? could does It does not have to be kind of clinical development or what have you. 06:37.35 Lana Feng All across this entire value chain, can we help them innovate, right? Without relying on IT t teams and IT typically will get involved typically. And then when they see, okay, this is seeing traction, right? They actually want to allow the SMEs to experiment. 06:56.19 Lana Feng We just didn't have the tools before, right? We didn't have the the resources. We couldn't hire enough software engineers to do this. So really this new wave in terms of Vibe coding, allowing kind of the local platform, the democratization of app building allowed us to do this. 07:12.51 Lana Feng So that our differentiation from... some of the other tools like you know the lovable the bolt and and and replit is that um we actually allow you to build ai native um application not just the front end right it's actually this production grade applications um without knowing how to write code But then, of course, you know, that the ah developers also love it because we've um we've done some ah kind of market research. 07:34.14 Ratnadeep Bhattacharjee Okay, 07:40.12 Lana Feng We've reached out to folks and we have a really long wait list. It's actually developers or tech adjacent are really interested in this. 07:48.72 Ratnadeep Bhattacharjee okay, no? So as you as you were kind of, yeah, it does, it does. 07:51.70 Lana Feng Does that make sense? 07:54.36 Ratnadeep Bhattacharjee Now that you, when you were explaining, when you were kind of explaining what you are doing and why you were doing it, one of the things that came up to my mind is is the is the buzzword, right? 08:01.02 Lana Feng Yeah, that's the why. 08:06.79 Ratnadeep Bhattacharjee Agentic AI is the buzzword now. I mean, you cannot just escape it. But according to you, from ah from ah a techno-functional person like you, how is this different from the traditional AI or even the ah current wave of gen AI tools like ChatGPT, for example, right? 08:24.73 Ratnadeep Bhattacharjee And why is this distinction important, do you think? 08:25.20 Lana Feng Oh, that is... 08:29.33 Lana Feng Okay, so um I'm so glad you asked this question. It is a buzzword and it is like actually a lot, very, very fuzzy, right? It's overused, but don't really know what it does. So um we're all familiar with kind of ChatTDT, right? You ask a question and then you get answers. 08:44.88 Lana Feng So very much kind of you ah obtain that information, but it's very kind of the iterative, right? Conversational process. Agents, on the other hand, are very different. but So think of them, they do two things. 08:57.03 Lana Feng First is they can autonomously perform tasks that require multi-steps. right So that's one thing, because actually does stuff. And secondly, it can orchestrate complex workflows. 09:13.70 Lana Feng So that's kind of the biggest difference. Yeah. So what we do is we actually take um a step further and saying, okay, can we actually, agents are really great, right? 09:16.28 Ratnadeep Bhattacharjee Okay, okay. 09:23.81 Lana Feng They can do something and um not just going to prevent the answers. And can we actually take this to a next level, right? If you think about MCPs, MCPs are basically these kind of, think about it as UCB, right? 09:37.82 Lana Feng The plugs, right, allows all of a sudden but the the devices, your Apple devices, your iPad, your phones, and your computer to be able to connect to each other. Or maybe HTTP for the web, right? You think of before, it was everything was custom. But once we adopted the HTTP um standard for the websites, all of a sudden, you can't access any websites. 09:57.80 Lana Feng So this is kind of that standard. So LLMs provide that information, right? And then you have world traditionally run APIs and software, right? Those are kind of battle tested. We know it works because it runs a world, right? 10:13.19 Lana Feng Data, what have you, but then they don't really connect, right? You build agents and improve workflow, automate some workflow, but they don't really connect. But then um what MCP now it does, it allows you to kind of connect these, right? 10:27.97 Lana Feng Think about you put a like MCP enabled on all the traditional tools. So you think about LLMs are kind of the brain, right? And then the MCP enabled MCP servers, we call them basically MCP tools are the bronze, right? Actually do things ah for the real world. 10:45.24 Lana Feng So the very first time we can connect those two. 10:49.75 Ratnadeep Bhattacharjee but That's a very interesting way to put it. you know well One of the things that I was also thinking well while you were explaining is how does this really kind of impact the life sciences ah and healthcare you know sector as a whole? right how does it How do you think it it it will kind of change the whole paradigm? 11:11.28 Lana Feng oh, it's gonna is really gonna transform the entire healthcare life sciences industry, right? Think of what we have. but have ah We have lots of data, right? We have really complex workflows, and then we have expert in the loop, right? These are the things that i must have for LLM. That's why initially we ran into some challenges in terms of adopting genitive AI, right? It's that where's the guardrail, where's the transparency, where's the accuracy? 11:37.17 Lana Feng The workflow is really kind of the key. This is where well suited for agentic development, right? Remember what I said, there they can autonomously do tasks that requires multiple steps and can orchestrate complex workflows. 11:51.98 Lana Feng So really, I think um Accenture came up with a research and saying just agentic AI development alone can basically yield $50 billion dollars in kind of economic impact. 12:04.97 Lana Feng in the next ah five years for healthcare life sciences. 12:07.60 Ratnadeep Bhattacharjee wow 12:08.46 Lana Feng If you think about it, and I'm going to start with the healthcare first. right It's about you know um back office automation. right Think about the complex workflows. Think about prior authorization. right Think about automating kind of co-pilot quote unquote for clinicians, reducing billing errors, right? 12:27.93 Lana Feng Reducing burnout, kind of inscribed, you would have seen really big companies, right? Already kind of making impact there. So on that side, right? So lots of like personalized treatment, really kind of think about, we need to transform healthcare care from more of a sick care to kind of outcome-based, right? 12:46.79 Lana Feng care, right? so So that is, I think, is a great, could be a great driver. On the life sciences side, we adopted genitive AI, large language models, actually quite early on, right? 12:52.71 Ratnadeep Bhattacharjee Oh really? 12:58.82 Ratnadeep Bhattacharjee oh really 12:59.11 Lana Feng Drug discovery. Yeah, drug discovery. If you think about, you know, the the the the models, we you know, we use um ah molecule models, alpha-fold and what have you to do kind of um protein design, right? 13:11.10 Lana Feng For drug discovery, now it's like, you know, um um ah adopting generative AI in a really big way. Like I said, from drug discovery all the way to to commercialization. 13:21.83 Lana Feng So here you can have, you know, um ah ah agents are really great for smart clinical trial um orchestration, right? Think about, you can do drug design, you can actually do smart screening, right? It's basically pretty much kind of every um fabric of um life sciences can be accelerated. 13:43.17 Lana Feng And that my apologies. So that's kind of um what I foresee and the impact it could have on healthcare life sciences. 13:55.03 Ratnadeep Bhattacharjee oh Okay, okay. So ah ah one of the things that that often comes up in these discussions, especially on AI and agentic AI, right? Especially now that we're talking about life sciences, right? Life sciences a domain, I believe, where the cost of error is high. Like high would be an understatement, right? 14:15.18 Ratnadeep Bhattacharjee How do you ensure ai outputs are not only accurate, but clinically responsible, if you know what I mean? 14:15.43 Lana Feng Right. 14:23.22 Lana Feng Right, yes, yes. So, um you know, models actually are getting a lot better, right? Like when we first started um working on the DaVinci models, I mean, we're talking about what 70%, 80% accuracy, right? 14:37.24 Lana Feng But then now it's getting to, you know, 90%, right? So, um i and then of course, there's the ah AI approach, identity approach, you could basically deploy evaluation evals, right, to kind of, um sort of this this ah blackhead approach to kind of be skeptical, right? That the answer is coming back from from your LLMs or your genetic system, is it correct? 14:59.94 Lana Feng That's the first thing is AI against AI, right? To kind of um ah ensure that the accuracy is there. And secondly, of course, you can actually build um humans in the loop, right? 15:11.27 Lana Feng To be able to kind of check every, and let AI you know be transparent show your work, and then have experts like check the work. ah So, um and and they involved every step of the decision making process. And then thirdly, this is very well unique. 15:28.40 Lana Feng to the agentic development is that um really kind of, I like to to say not human human in the loop, but human on the loop, right? So from kind of humans doing some decision making in the loop and doing some of the work from this whole process to agents do this work, but humans provide kind of the oversight. It's almost like you're the manager of your digital agents. 15:52.84 Lana Feng ah 15:55.77 Ratnadeep Bhattacharjee No, no, you're you're right. You're right. So ah one of the important things that you keep mentioning and emphasizing is, you know, expert in the loop, human in the loop. So this is a model, right? 16:07.09 Ratnadeep Bhattacharjee Typically within the agentic AI ecosystem anyway, right? How does that work in your systems? And how do you maintain the balance between, let's say, automation and ah expert oversight? 16:19.68 Lana Feng Oh, because we came from our previous background. So, like you know, ah accuracy, transparency and human in the loop are really, really critical for us. For even Coxia, what we're building is that every step there is a checkpoint. It's like, you know, how do you want to proceed? Right. So first thing is transparency. Everything is streamed on your desktop and showing what the AI is doing. 16:42.91 Lana Feng so that you can actually check, right? Because it's actually giving you what are the components you're using, what is the accuracy rate, and we actually build kind of a system where you can actually evaluate the components. Are they really good? 16:58.17 Lana Feng Are they sourced from reputable like sources and all these kind of checks and balances? transparency that we can we can put it in there. We also put in observability, right? It's like you'll be able to see what is going on there. 17:10.30 Lana Feng And then thirdly, every step has a human to kind of check, right? It's like, do you want to proceed? Do you want to modify? i think eventually, based on kind of user behavior, because it you know, chat GPT has, where it kind of is widely adopted now. It's like what I heard last was like a billion users. It's like astronomical that once you kind of get comfortable with the system, like you're comfortable, let the system building your application, AI native applications, you're comfortable with the quality, you're comfortable with the sort of the transparency and accuracy, you will kind of go in less in terms of checking every step, but it's there for you to check. 17:50.01 Ratnadeep Bhattacharjee Yeah, absolutely. Absolutely. 17:51.71 Lana Feng Sorry, it's a long-winded answer. 17:51.53 Ratnadeep Bhattacharjee You know, ah no, no, but this is, this is very important, you know, because as, as, as you, as you can understand, right? 18:01.92 Ratnadeep Bhattacharjee Because this is an interesting segue to my next kind of question that I, that I, that I, that I constantly keep you know, thinking about, right. You know, one of the most powerful ideas in not just your work, but you know, most of the, uh, companies which are building AI products or building solutions on top of AI or, or leveraging ai is the democratization of AI development, right. 18:28.03 Lana Feng Absolutely. That's what we're aiming for. 18:28.92 Ratnadeep Bhattacharjee Uh, Exactly. So how do you how do you kind of enable life sciences professionals, right? People without a technical background, right? To build and deploy production-grade AI workflows. 18:43.51 Lana Feng I really see this as as a teamwork. It's not just like one lone wolf will be able, but if someone who's like an early adopter really tax that tech savvy, like for example, a data scientist and what have you, right? 18:56.49 Lana Feng You definitely be able to do this. The democratization is very kind of, um wide in terms of interpretation, right? You have to be kind of have loft missions, but you also be realistic, right? Who would be the early adopters? 19:12.06 Lana Feng Right. So that's why we're saying tech adjacent. So, um but the the long story short, bottom line is that anyone with an idea, right, will be able to use our platform. 19:23.41 Lana Feng You don't need to be able to write like Python code and what have you. 19:23.98 Ratnadeep Bhattacharjee Okay. 19:27.29 Ratnadeep Bhattacharjee Right, right. So you use the word tech adjacent, right? Can you can you kind of explain it a little bit more? 19:34.82 Lana Feng Oh, ah it could be, um you know, it could be like a maybe an IT t business partner in a particular like business unit, right? It could be, you know, these um in pharma, particularly a lot of these innovation folks actually came from tech. 19:51.36 Lana Feng So they kind of have dabbling, they like, you know, command line, right? They write SQL queries and what have you, particularly data scientists, right? They do write some code. They use Python very heavily, but but they're not software engineers, right? 20:04.19 Lana Feng So, or even like, you know, yeah, exactly. 20:04.85 Ratnadeep Bhattacharjee Correct. Correct. Correct. No, I agree. but and But what kind of shift have you personally seen in how these, not data scientists per se, but like non-tech professionals, right? 20:20.72 Ratnadeep Bhattacharjee What kind of shift have you seen in these non-tech professionals think about i mean and, you know, how they interact and think about data and automation? Have you seen a shift? 20:30.67 Lana Feng Oh, we're definitely seeing the shift, right? The shift came after ChatTPT was adopted, right? Because I remember we were at um our previous company, we actually have um our our customers are like, they're using ChatTPT and saying, hey, you know, how are you different, right? 20:47.33 Lana Feng So most the companies that we talked to in life sciences had deployed some kind of a private version of ChatTPT. so that they could access, they could ask questions of their internal data and what have you. So that is a very, very white widely adopted in terms of AI applications. 21:03.81 Lana Feng And secondly, is that how do they kind of see the business challenges they can leverage um a Coxia for? So I think this is where um you know experimentation starts small. 21:16.33 Lana Feng So good news, we're seeing this kind of democratization, um success of these democratization platforms is like you know the livables and the... 21:20.61 Ratnadeep Bhattacharjee Got it. 21:25.81 Lana Feng right It could be like you know someone... In the healthcare care life sciences, they probably created a small website for their hobby or what whatever. You know what I mean? because i We saw this parallel in ChatTPT as well for our customers. 21:39.19 Lana Feng They were using ChatTPT to kind of for the kids to do their homework and what have you. 21:39.11 Ratnadeep Bhattacharjee that 21:44.36 Lana Feng They're like, oh, I can use this for my work as well. So we definitely see that very wide adoption for on the consumer side for these kind of web application builders that are democratized in a low-code, no-code platform. 21:58.84 Lana Feng So we're quite bullish about how the reason that they can, I don't know whether you've used these platforms before, is that, um you know, it gets you like 70, 80% there. 21:59.65 Ratnadeep Bhattacharjee Right. 22:08.41 Lana Feng You're like, wow, the first version is there, but that's not exactly what what I need, right? 22:09.94 Ratnadeep Bhattacharjee Yeah. 22:12.57 Lana Feng I need to go in and finish that 20%. 22:12.55 Ratnadeep Bhattacharjee Correct. 22:15.21 Lana Feng That is really, really, really hard. It's like, because you can't rely um entirely on the models. So this is kind of where we come in. It's like making it but product production grade so that you can actually use the app to do something, to collect with your data, to do the analysis. 22:31.63 Ratnadeep Bhattacharjee That is, I think one of the important terms you use is production ready, right? 22:35.31 Lana Feng Yes, production ready. 22:35.33 Ratnadeep Bhattacharjee Yeah. yeah i'm okay I mean, a common challenge that you see, especially in enterprises is the POC trap, right? Many AI tools get stuck in pilot mode, right? 22:44.45 Lana Feng Yes. 22:48.49 Lana Feng We call it pilot progatory. 22:48.26 Ratnadeep Bhattacharjee What lessons... 22:52.40 Ratnadeep Bhattacharjee ah Really, that that's a very good way of putting it. You know, ah I always had these questions, right, especially for people, someone like you, right, a veteran in this space, right? What lessons can you share about moving from experimentation to enterprise adoption, especially in, ah you know, regulated and highly complex environments like probably biopharma or life sciences? 23:16.34 Lana Feng Oh, I'm still learning. it's It's, you know, market is really complex. I can only kind of um share some of my personal experiences. I really think um um um one key advice I can give is really make your product like really, really good, right? is um It's so anyone can use it. 23:39.81 Lana Feng So it's just spend the time, build a product and then test it out. 23:46.28 Ratnadeep Bhattacharjee Yeah, I think it's pretty simple. 23:48.64 Lana Feng Yeah. 23:48.73 Ratnadeep Bhattacharjee When you put it this way, it looks like, you know, the purgatory really is a man-made, you know, mental purgatory. 23:57.32 Lana Feng Yeah. 23:57.94 Ratnadeep Bhattacharjee So... ah 23:59.06 Lana Feng Yeah, purgatory is the right word, right? It's a, you know, we had all, all of the startups have had the same experience. just pilots after pilots. Yeah. 24:08.39 Ratnadeep Bhattacharjee Yeah, exactly. You know, even we we face the same situation with many of our enterprise customers. So, you know, I understand we are in the same ship. 24:15.52 Lana Feng And then and part of the, I kind of flipped this, right? 24:16.28 Ratnadeep Bhattacharjee ah 24:18.72 Lana Feng Maybe the reason we stay in pilot is like we're not offering enough value, right? 24:23.77 Ratnadeep Bhattacharjee Maybe, yeah. 24:24.51 Lana Feng We see quite a bit is that they would, you know, pharma companies, they would experiment and then they ended up learning and they ended up building something themselves, right? That really kind of tailored and customized to their needs. 24:36.55 Ratnadeep Bhattacharjee Absolutely. Absolutely. No, this is a very good point. and I agree. ah um Moving on, right, you know, from early discovery to post market surveillance, right? 24:49.16 Ratnadeep Bhattacharjee AI is kind of touching every stage of the life sciences journey. 24:52.74 Lana Feng Correct. 24:53.29 Ratnadeep Bhattacharjee Where do you think agentic AI will make the most impact in the next 12 to 24 months? Are there parts of the pipeline where it's still early and so and areas where it has gone on to a certain level of maturity? 25:09.53 Lana Feng I'm thinking where is kind of the workflow is the most cumbersome, right? um I think probably clinical trials comes to mind, but on clinical trials is such a ball of wax, it's like so complex, right? 25:22.99 Lana Feng And then the solution market is so fragmented, it's going to be really difficult to tackle. 25:24.13 Ratnadeep Bhattacharjee Oh yes. 25:27.47 Lana Feng But I would say kind of, you know, the orchestration of clinical trials, for example, and then we're seeing early success in drug discovery right so maybe in silico screening um it's almost like every single so space within this entire value chain will be um um sort of impacted and early success you know customers customer customer service right customer success anything maybe on the sales side we're seeing early adoption there as well so it's of course it's always commercial they're they're the most open 25:32.53 Ratnadeep Bhattacharjee Okay. 25:51.59 Ratnadeep Bhattacharjee Hmm. 26:00.90 Lana Feng to innovation, right, because it's the bottom line. So um so we're seeing early successes there. 26:04.48 Ratnadeep Bhattacharjee right 26:07.43 Ratnadeep Bhattacharjee So on the top of your head, you just mentioned clinical trials as one of the use cases where Asian TKI might potentially have a big impact right or is already having a big impact. right can Can you explain it a little bit more, you know, in terms of the process of clinical trials and where exactly i mean within the workflow agent TKI will really, really have an impact? 26:32.95 Lana Feng um You know, patient recruitment, right? If you think about it, the whole process of um finding patients and getting them enrolled, right? And then getting them to the to the clinical hospitals where they're enrolled in the clinical trials and then the visits and then all the testing and what have you. 26:53.07 Lana Feng So those are anything that has kind of really, really complex processes, right? And involves lot of stakeholders. can be somewhat automated. A lot of the back office stuff at the clinicians, clinic, clinician offices that involved in clinical trials can also be an automated, right? it's It's not completely automated. Like I said, like the expert in on the loop or expert in the loop to really kind of provide that oversight. 27:19.72 Ratnadeep Bhattacharjee Okay. Okay. Uh, I mean, a very hypothetical question. ah do you imagine a world where, uh, AI agents become true collaborators in scientific workflows, you know, not just assistants as they are right now. 27:37.33 Lana Feng um I definitely see that in maybe five, 10 years, maybe even sooner, right? As the technology evolving so quickly that it will be able to kind of in a true kind of co-pilot situation where, you know, like scientific literature review, we have lots of AI solutions, right? 27:43.80 Ratnadeep Bhattacharjee Okay. Mm-hmm. 27:55.16 Lana Feng It help you doing that, at like post-market surveillance and what have you, right? And how do you really kind of as the value increases, and I think it's also the adoption and the trust. 27:59.80 Ratnadeep Bhattacharjee Okay. right. 28:05.80 Lana Feng is the adoption, like how this this is this the entire ecosystem, right? How do we actually prepare our workforce to be kind of, um we're not talking about AI native, we're talking about kind of the open to ai solutions, right? Because we saw a lot of resistance before where, you know, trust and also, um you know, the concern about replacement, right? 28:29.82 Lana Feng So really kind of how do we upscale and reskill our workforce to be kind of AI ready, right? And then the next thing is, of course, how do we make the data ready? 28:37.62 Ratnadeep Bhattacharjee good 28:40.72 Lana Feng ah You're talking about governance, you're talking about ah be the the data cleansing and standardization, right? 28:41.41 Ratnadeep Bhattacharjee Yeah. 28:46.95 Lana Feng There's all these, all the pieces are going to be involved to actually for us to get there. 28:52.61 Ratnadeep Bhattacharjee No, I absolutely agree. you know, AI Retiners... before that even data readiness and before that even you know tech readiness you know all of this should ah be in place to make sure that we are there at a certain point of time so yeah some people will get it sooner and some people will eventually get it right so i agree with you 29:12.48 Lana Feng Yeah, we will always have early adopters and later adopters at the bell ah classic bell curve. 29:14.02 Ratnadeep Bhattacharjee ah 29:17.19 Ratnadeep Bhattacharjee exactly exactly i agree yes yes the classical curve i agree um Lana, before we close, right, what advice would you give to healthcare care innovators, digital health founders, or even enterprise leaders who want to build, you know, ethical, effective and future ready AI solutions? 29:41.00 Lana Feng I want to say maybe keep an open mind, right? and Start and like experiment because it's pick some of the low hanging fruit use cases and start small. and And also, you know, prepare the workforce to um to kind of, because you don't want to spend a lot of money in in piloting and then find out like you're, The team, internal team, actually super resistant to this. Engage with your stakeholders early to really truly build these kind of a holistic approach to adopting AI. 30:15.32 Ratnadeep Bhattacharjee I think AI readiness, you know, that is that is what it starts, you know, people readiness, then, you know, you can move towards readiness. 30:21.54 Lana Feng And make the investment. 30:23.37 Ratnadeep Bhattacharjee Yes, exactly. That's the most important part. I agree. 30:26.39 Lana Feng Yeah. 30:28.29 Ratnadeep Bhattacharjee So, ah Dr. Fang, you know, thank you so much for this deeply enlightening conversation. Your work is an incredible you know example of how AI can be harnessed, not just for automation, but for empowerment and innovation in healthcare, right? 30:44.52 Lana Feng yeah 30:44.61 Ratnadeep Bhattacharjee To our listeners, if you found today's episode insightful, please share it, subscribe it and stay tuned for more conversations that blend AI, clinical excellence and system design. 30:55.34 Ratnadeep Bhattacharjee Dr. Lana, where can people follow your work or learn more about what you're building? 31:01.88 Lana Feng Just go to our website. 31:02.19 Ratnadeep Bhattacharjee What's the best way to reach you? 31:03.59 Lana Feng yet Just the website on cogzia.com as in cognition, C-O-G-Z-I-A. 31:13.16 Ratnadeep Bhattacharjee Thank you. Okay, one second. Thank you. And to everyone tuning in, this is Ratna Deep and you have been listening to Leaders Perspective. Until next time, stay inspired and stay intentional and inspirational both. 31:26.85 Ratnadeep Bhattacharjee Thank you. 31:28.06 Lana Feng Thank you so much. Bye.