0:01Hey, guys. Welcome to the podcast. Have a really exciting, kind of almost a special episode to share with you guys because this is a part two of a prior episode. So I wanna introduce and welcome Nandanay and Vivek to this episode, which is gonna be called the augmented MSL part two. So, Nandanay, I'm gonna have you introduce yourself first, and then we're gonna get into kinda how this whole thing came about.
0:33Hello to the MSL and Medifference community. I'm Nandini Sabrawal. I work as a field medical director on the internal medicine team at Pfizer. That title translates over to an MSL. I have redoubt my title to the augmented or the AI MSL. Just as a disclaimer, the all opinions are my own. They nowhere reflect any of the company's stance on AI or anything for that matter. I'm happy to bring a legend to you guys, one of my greatest collaborators, Vivek.
1:07I will give you I'll pass it on to you. No. Thank you, Nandini. You're too kind. And, again, it's incredible. And I was just you know, Tom, it's been it's just an incredible pleasure to be on your podcast. This is I I watched them. I've I've I've kind of cherished everything you've done for the for the community, and this is just an incredible opportunity to contribute. So as Nandini mentioned, my name is Vivek Mukathyar. I am the medical AI team lead at Pfizer. It's an evolution of my team now that has started about a year ago, my journey at Pfizer.
1:41My background's in medical affairs. I have spent time in the field. I've spent time in house. I've spent time in in pipeline medical affairs roles. And every type of medical affairs function, I've had some exposure to. But, of course, my my near and dear has been the time I've spent in the field really engaging HCPs. And now I get to do that by incorporating AI. So that's my role really is around how do we leverage AI to augment medical affairs function, including HCP engagement where MSL work falls.
2:13So incredibly excited to be here on the podcast, Tom and Nandini, and thank you for the invitation. Well, I guys, I'm I can't tell you how thrilled I am to have these x literally, AI experts. These guys are doing things that nobody else is doing with AI in the MSL community. So I'm so excited to share them with you. Let me tell you how this came about. So first of all, Nandanay and I did an episode a year ago. It was October 15 called the augmented MSL, which is the, you know, the first episode we
2:44did on the AIMSL. She crushed it, of course. And it was amazing. A lot of people commented and got a lot out of it. And then ran into her literally this week at Fierce Pharma Week, which was awesome, by the way. Great show. Yesterday, we came up with this idea to do a part actually, Nandanay came up with the idea to do a part two. And here we are. Literally, within twenty four hours, we put this whole thing together. We're presenting to you part two because we want to show what's changed within the last
3:22year in AI. But before we do that, Nandan, I'd love for you to do a recap of what we talked about a year ago so you can catch everybody up to speed so that as we can get transitioned to, you know, this new stuff, they'll be caught up. The so from a year ago, we talked about how to go about learning prompt engineering. We define what it was and how to go about upskilling yourself. And then how to get leadership buy in because to get access to these GenAI tools.
3:58We looked at a few use cases where it was based on role playing interactions with health care providers. It was learning content if you are new to the therapeutic area, and then how to use ChatGPT to write effective emails as opposed to the generic emails that it gives to us. And then we looked at some of the limitations of these AI tools a year ago. So fast forward a year later, a lot has changed. And really, if you wanna be able to, for this episode what we'll focus on is if you really
4:33want to get to the next level, you know, how do you go about doing that? And as an MSL, that role is still it's still an an individual contributor, and then it takes a while to build skills to work in higher impact things. And then so you really need to see others in your organization when it comes to AI, and you have to be a leader. And that's how I start out, Vivek. Vivek and I, we met this was the backstory.
5:05Even we Vivek and I, we we met back in January, and then we realized we were both speakers at the same at a conference on AI and medical affairs. And then so we met, we built a relationship, and so I made the decision to put reoccurring meetings on Vivek's calendar. And then so I so it I held them to that, and we would I would create discussion topics when we would meet.
5:36And then if Vivek and I if Vivek and I, we we would from there, it's it's just having an conversation where Vivek, we would just he was like, hey. This is what I'm working on. This is an app that I build. You wanna check it out? And then I would check out his app, and then I would learn from there. And then so I it was I slowly built my skills. So we just it was a we kept inspiring each other. So every time we would have our meetings, we would walk away super pumped, and that would give us energy to work on the next big thing.
6:09Yeah. And I'll just add to that, Nandini. Again, you've been you know, this is a fantastic summary of where we were. But, again, I think I just double down on what Nandini mentioned that how I think the way to learn in this world is to really do it through collaboration. We are all learning new capabilities. The new capabilities get in get incorporated into our our tools that we have access to, and and I think that's that's the key piece that Nandini brought up. That it's just a collaboration, really sharing our best ideas, and and and then kind of let it
6:40our own creativity and imagination is now these tools allow us to do everything. So, so that's been great. Did did you guys meet when you say January, was it at the AI conference in Philly? Is that where you met? Just right before that, we saw that we were on the agenda and we worked for the same company. So I was like, well, let's at least align on what we were going to say first. And we didn't know that we were all independently working. But since then, we we do nothing at this point. We've been we've been we've been exchanging ideas, collaborating, and really pushing the
7:12boundaries and frontiers of what can be possible in the world of, you know, MSL and AI. That's amazing. I love that. So I by the way, I was there, so I'm gonna take full credit for the whole reunion. 100%. But I love the fact that you you guys kinda sought each other out. And I know, Nandadeh, it sounds like, you know, you really were proactive in in making sure that you positioned yourself in Vivek's, you know, on his calendar and and got that that those regular meetings.
7:43And you guys are learning from each other, and you developed and created this mentorship. And that's how it's done, guys. I mean, that that's how it's done. This it's so powerful when you collaborate with other people that are that have great ideas, that are that are doing innovative things. So I totally encourage that, but I I'm curious now. So, Nandade, as you're, you know, a self proclaimed AI, augmented MSL, and, Vivek,
8:15you're a gen I Gen AI lead. How do you guys work together? Yeah. So let me let me let me take that on from that perspective. So anything I'm gonna say right now is, like, it's my own perspective and not that of my employer. Right? With that in mind, I think when I look at, I'll go back about couple years ago when I was I was still in medical affairs. I was a medical director. And and in in that scenario, how I would have worked with you know, I was doing the same
8:46thing. I was reaching out and working on ideas and the way we collaborate as well. That Nandini has an idea. I think she's a superstar. She's coming with a lot of great great ideas. That we will say, hey. What is the where is the technology right now, and what can we do with it? Let's create some prototypes. Let's play with it until we see and then just understand where the boundaries are. So that's where how I approach to it. But, Nandini, you can please chime in with your perspective. You know? Do yep.
9:17And for it's yeah. I was working towards up skilling myself, and then there's I I was heel hitting the ceiling. There was I had to find others outside of my team who who were a a few levels above me. And then Sveveig was one of those persons where I could bounce ideas because then I was learning, we were learning how we were speaking the same AI language. And so whenever I would have, you know, these ideas and then, to really take it to the next
9:47level, that is a heavy lift because as MSLs, a lot of our time that we spend is in the field. And then to add on an additional project or an additional skill set, it's time that you have to you have to figure out how to do that with your time. And then so, yes, I was listening to all these podcasts and AIs when I was doing my drives to health care provider offices and stuff, but then now came the time like, okay. I have all these ideas. How do I bring it to life and execute it? And then so Vivek was helping me do some of that things.
10:19So I had to learn skills, for example, Vivek, how do I structure my ideas? How do I pitch it to the leadership, my senior leadership team? And then he would, I would show him like, oh, these are the slides that I created for my project idea. And then so if I'm able to review it and he would say, know, not that you need to add in an impact slide. Like, what is the bottom line? How much, you know, time saved are you gonna have by using by going to manual route versus using using a Genie tool? And then so it'll be, like, little things like that that Vivek would give me a lot of feedback
10:52on. And then he would share some of the content that he was creating, and then so I was building skills on top of that. So you definitely need somebody who to to collaborate with. Because it's going because of the nature of the role, you're just not gonna be able to do it yourself, and you're gonna be stuck on just doing use cases for yourself. Mhmm. But the whole goal of AI is to uplift use it to uplift yourself and your team. And then so having, you know, having a team like Vivek's is instrumentally extremely
11:27helpful. It will really will help you get to the next level. And I think that's what people wanna know. People are like, how like, you you hear about prompt engineering, and it's like, okay. So I'm gonna start to get really good at my prompts. But a lot of people fail in not realizing or recognizing how else they can be using it. So the use cases. So I know we talked about that a year ago. We talked about certain use cases. We talked about prompt engineering.
11:59Catch us up to speed. Like, in the past year, let's talk about what's evolved, and let's talk about how things have changed in that year as it relates to AI. Me. Vivek, do wanna weigh in first and then I can go? No. Please. Go ahead. Go ahead, Nandini. What has so after I, you know, gained a really good foundation of prompt engineering, and it's just a way of communicating with language models. And from there, I went on to doing like vibe coding came onto the scene where, you know, you
12:35can just interact with apps and platforms like Cursor and Replitz. And Vivek was introduced, he had introduced that to me. And then in tandem became the rise of the agents. And I I mean, that's a good thing. That's a good I'm saying in a good way. And then so it was you know, you still need to know the like, have those foundational skills of how to effectively communicate with the language model and telling it, you know, exactly what you wanted to do. And then so when I started out on these on the route of an agent, then I would get counsel
13:10from Vivek. Yeah. And I'll I'll just add to that, Nandini. I think Nandini is shelling herself short. I think the intention of wanting to learn and to grow this additional skill is so critical in this in this environment. And and to her point, right, point engineering, I think folks folks, I think, would appreciate, right, that having the ability to have the someone to be able to speak the same language. Right? We are all brand brand new to this the whole whole new vocabulary of AI and LLMs and prompt
13:43engineering, white coding, context engineering. That's the kind of evolution of prompt engineering, which, you know, we can get into that at some point as well. But understanding how the field is evolving and why that leads to that, hey. You know, that's gonna unlock new use cases. So for your viewers, if you have not you know, because you had a bad experience with an LLM, and then you just alluded to this, that how do you talk to the AI is what what initially you need to learn. But what you will find out that how AI response evolves every single day potentially or at
14:17least in a big step jump when you go from a g p d four o to the g p d five, which was recently released, your responses are significantly better. And and even though your prompt is the same, the way the way the intent is understood is a lot better. So there's constant, and it's and and and it's a unique part of human history where we are in, that the technology advances are exponential. So to be able to keep up with it, to have partners, allies in the organization that'll help you inspire you, like Nandini doesn't do every every time I speak to her, then who's
14:50excited to learn. And then it it kinda motivates me to come up with, like, oh, you know what? Can I push the boundaries? Can I solve this problem for her? Can I find some unique use cases, like, unique applications, and can we build something? So so that's sort of, like it's a it's it's it's this common language we need to start speaking as a as a organization. Yeah. And I think one go ahead, Tom. Go ahead. One one of the best ways of of learning and of of, like, of teaching this this particular
15:24skill is to show examples on how to go about doing this. I knew, like, prompt engineering was a very scary word in the beginning, But I would say, like, we've moved, you know, moved away from that where you don't have to necessarily have this you know, you have to understand it and understand kind of a basic framework and what you want from the from the LLM. But there's lots of different ways of going about doing that. And I'll show you, I'll share my screen and I'll show you some examples of how to go about
15:54doing that. So it's kind of like, you know, you have to say, how do you take the idea in your mind and then, you know, make it come to life? And this is an example of how to do it. So, guys, anybody that's listening right now, don't crash your car, like, if you're driving or if you're at the gym. You may wanna check this out on YouTube so that you can see the visual. K. Let me move this out.
16:24So, say, this is this is, you have an idea, and this is a very common one. I'm using my own CHAD GPT-eight personal account and here's what I told you, CHAD GPT-five. I'm working on a project for pre co planning for my field medical team. The main goals of the project are listed below, and your task is to help me structure this project, give me an overview that outlines the core pages and features of this project with example use cases.
16:56And then so for this one, I just had to focus on one part of the project. So say I wanna do something like provider profiling. So it's other words, give me a brief snapshot of the health care providers and then, create a profile that focuses on their shows me a research, recent public appearances, any recent publications that they've done or previous conferences that they have attended. And then so I went ahead and I just did this about fifteen minutes ago.
17:29But then so you can see some of how it's I wanted to show you how it's structuring the project for you. So it's doing a pre call planning project overview. It's doing a provider dashboard, and it tells you what the core it's, you know, creating the core elements of this project where it's showing you, okay. Here's the all basic elements. It's gonna give you basic information on the health care provider. It'll give you the research focus, publications, it's going to tell you what it's pulling in from there, it'll give you the title, the journal, a short summary of the
18:03publication, public appearances, past engagement histories, any network map if they're connected to other KOLs or institutions. So look in the public domain and try to find this information. It's there's a call objective planner. There is insights and note capture. It can do conference and events tracker. So it's building all of these. It's building on top of the main ideas I gave it for my project.
18:34And that's the that's the value in it. I'm not I didn't do any prompting. I just told the the ChargePT, you know, what what I kinda wanted to work on. And then so the next step is, like, okay. If you wanted to see what the prompt actually would look like behind these modules, so then you tell it, generate the key AI prompt and the task specific prompt for each of these modules. And then so here, it's it'll give you the system AI prompt.
19:06So it only writes it out for you. I never have to go in and do all this work. I can go in and modify it and tell it to change certain things around based on how I want to see it, but it just does it. It creates a task specific prompt for the provider profiling module part of it. The call objective planner, it creates, here is the task specific prompts. And then so it goes, it continues on for each of the modules. And I am like, okay, this is all great. What do I do with this with things like this?
19:39So this is an example of how you would see detailed prompts, but then what do you do with it? You have that, you save that, and then you can continue the conversation with CHARGPT and say, hey draft me a sample of a wireframe or a page flow. So, an outline and simple visualizations so you can show what this your team, what this would look like. So it's it's grabbing all of the research it's done on the project overview and the prompts
20:14itself, and then it's creating it's creating an outline for you. And then so here, it's, like, gives you a page one. It gives you a very, like, nice, easy, visual depiction of what it would look like for the providing profiling part of it where it's showing you, like, okay. Here's your profiles, there's provider snapshots, here's what you would see it as a glance. And then it continues onwards into a more, very more like cleaner, simplistic visualization of it. And then so all this information you can gather together, and it will help you create a a a
20:51pitch deck that you can use when you're pitching to your your leadership and saying, like, hey. This is why this is where, you know, I have this project idea, and then this is what the idea is, and where do I wanna take it. So really simple. Like, you can have it develop these simple proof of concepts for you. And that's kinda how you take it to the next level. I love that. Guys, like, I it may have been hard to follow if you were listening, but you really wanna check it out because the whole idea with AI is to see what good looks like.
21:28What is a good prompt? What's a thorough prompt? How do you encourage the prompt? How do you take the this from point a to point z through continually prompting? I mean, am I saying that right? Yeah. That's a good sound. Yeah. So I think how do you talk to I think what Nandini folks should appreciate is that how is Nandini talking, and how is she verbalizing things and getting the response from the AI,
21:58and and then modifying AI's direction. So think of an AI tool. Like, you know, if you have Copilot at work, you have at work, you have Anthropic, whatever you have at work, think think of it as a direct report. And that is extremely, extremely smart. And and you start instructing it. And and the great thing about this direct report is that if you don't know what to do, you can ask it. Like, how can you help me? And it will and I'm gonna use the word it or he or she or whatever pronoun you wanna use, but
22:30it will come back and tell you. Especially g p d five is incredibly, incredibly powerful where you can you can say, think hard, but but also help me develop this prompt or help me help me how can you help me? Tell tell me how you'll help me, and it's it's incredible. So And that's an example of how I use it as a thinking partner when I really wanna take things to the next level and develop, like, projects that can help my team. That is one way of going about it.
23:00And then so, like, all the work that I do, then I chair with Vivek and say, Vivek, what do you think about this? Like, how can I do you have any feedback? How can I improve this? And then, you know, how do I get more of a clearer message across my senior leadership team? Because you need the you need the blessings from them to go ahead and work on, like, work on these projects. And they do take time, but there are so many different So think about in your MSL, your workflow, when there's lots of pain points along that.
23:34And then so think about, okay, if there's prioritize your pain points and then, okay, if there's a pain point of, like, okay, I'm doing too much manual work with this specific specific task. Right? So now I'm like, okay. How can AI help me? I'm not gonna sit there for, like, and then be clicking through things for hours and, like I mean, I've gone past and gone gotten frustrated with doing too much manual work. So that and that's where AI can help you to reduce the amount of time that you spend on manual work. And that's something for provider profiling where we were going in and we're doing Google
24:06searches and we're using different sources to compile profiles for provider. You can have AI help you do something like that. And then you can structure it and then really help take it to the next level and then help your team learn the skill as well. Yeah. The the one thing I wanna go I wanna add and go back to something Vivek said. So at fear one of the couple of things that I took away from the Fierce Pharma Week sessions and shout out to Petrina Palette who did an amazing job with her AI session.
24:38That's where Nandini was. She was on a panel. And they were talking a lot about, prompt engineering. And Vivek just mentioned the concept of telling it to tell you what to do. Meaning, you can ask you can put in a prompt, but in the prompts, ask GPT, whichever you're whatever LLM you're using, to create the prompt for you. This is what I'm looking to do. Create the perfect prompt because this is what I'm trying to accomplish.
25:11And I'm like, jeez, that's brilliant. You you ask it to give you the prompt. Don't even rely on yourself to come up with the right prompt. Ask it to give you the prompt. The other tip that I picked up on, one of the one of the the panel members had suggested to try to get it to argue with itself. Meaning, you wanna get hear about, let's just say, the some type of opinion on something.
25:46But then you can ask it, okay. Now give me the counter opinion. So you wanna see it debate itself so that you can educate yourself on both sides of a situation if you're if that's what you're looking to do. I thought that was brilliant as well because it gives you the ability to continue to learn but not in a one dimensional way. You wanna make it as multidimensional as possible. So so, guys, I wanna get back to and one of the things that that I think is just really
26:19valuable in these conversations is to share additional use cases. So is there anything else that you guys wanna share that, may be different from what we talked about last time? See, I can add to that, that scenario you were just talking about, Tom, where it's a lot of fun, especially if you are newer to the therapeutic area, and you really want to understand the mindset of different ACP personas.
26:49And then, so what it is, the concept is running a business war game scenario. So what you can say is that, I mean of course, don't put this information into public language models and stuff. I mean if your company has a GenAI tool, use that GenAI tool because then you're putting your company's drug information in there. And then so you can say like, hey, let's just make this up, that there is a new drug therapy
27:19that came on for atopic dermatitis that came to the market six months ago. On one side of the table, there is an HCP who is a full adopter of this therapy. And on the other side is an HCP who is completely skeptic, who's cynical about it, and will never use the product. Create a dialogue between these two ACP as they battle it out and argue their viewpoints. And what the Al Anama will do will create these scenarios where the two personas of the ATP, the adopter and the cynical ATP, and then argue their viewpoints together.
27:52You could, of course, add data points in there about the safety and the efficacy of the products and the endpoints and then have them argue with each other, and it's really fun to see. And that's kind of how, you know, when you go in and and you kinda go in and you ask about how how to do something like that. So it's a really fun way to learn. You know? It's kind of like running an ad board scenario when you're getting feedback from providers, and it helps you, like, you know, really think about things in different dimensions.
28:22Yeah. You know, I I'm looking back and putting myself in the shoes of my MSL, like, me being an MSL ten years ago. Right? So I used to go sit on back like, you know, go present at back committees. I used to present we HCPs, nurse practitioners throughout the entire chain of folks. They used to be in, you know, surgical products. So we'd have to engage a lot of different types of personas. And I think this is an evolution. Right? And we're getting to this very personalized engagements with our HCPs. And that's where AI is incredible at.
28:53Right? Be able to have a single message that then can be communicated into 10 different ways depending on the stakeholders they're engaging. This is an incredible I think if I had that, right, I would have been even more effective in my interactions with my HCPs. But that said that I think the other salient point that Nandini brought up is if I was an MSL today, I wanted to learn the topic. I mean, this deep research tool. And then a big shout out to Katrina. She and I had done this podcast on this topic before, but this is one of my still my favorite
29:26use cases. Just throw the information, therapeutic area. I know I'm still new at my new organizations. I run into new therapeutic areas that I have to work with those collaborators. I was like, let me find out what's going on. Give me the landscape, and it will go spend half an hour, forty five minutes on its own. And that that's your evolution. Right? Non multi agent systems, agents, compute, all of these new level you know, things technology that has evolved in the last three to five months has enabled all these new use cases that will allow you into a point.
29:57And and, you know, I'm putting it a little bit into the the what's what's out of the possible today. If you go to ChatGPT and agent mode, just ask it to go and review your LinkedIn. Right? Just go it'll log in to your LinkedIn, click through each and every one of your post. But do you think they can evolve that into an HCP interaction? Like, find out where my what my HCP's are doing on the social media. Give me a summary of each of them. It will go spend half an hour, forty minutes. And in some cases, if we have the very expensive, which I do become, which I don't, it will go spend a whole day doing that work and come back and report you that this is my your
30:31profile. Imagine how much into how much more critical engagement you can, you know, you can you can develop from that kind of information and insights. So when you say I want you to explain something. When you say go in in agent mode, can you explain what that means? Because I think that that's one of the shifts that we've seen in the in the past year is that when we talked last year, it was just ChatGPT four, and it was you put in prompts, and then,
31:04you know, you got your responses. And now there's deep research. There's agent mode. Can you talk a little bit about what you just meant by that? Yeah. Absolutely. So I think we get to kind of go down in why this is happening. So there is a parallel shift that is happening in the AI world where the cost of compute the cost of calculations. Every time you put in a request in the chat, GBT is, you know, what is the color of the sky? There is a cost associated with it. It's because the there's a calculation that happens. And and on the servers, you hear about energy and servers and NVIDIA chips and, you know, all
31:36of these things. Right? So that cost of that query has gone down 40 x in the just in the last year. So imagine how much intelligence you have on tap. So when you think about talk about deep research, you you know, what we could have not done a year ago because it would have been very cost prohibitive is absolutely possible today. So that you can have the AI go on its own, go through 50 different websites, read each and every one of those websites, summarize it, and not just summarize it, and also extract
32:07insights from there. Right? So let's say, you know, I used to work in hernia repair, and I wanted to understand what were the latest dosing paradigm for for this drug in the hernia repair space in the last six months. That's the only prompt I'm gonna put in. The deep research tool will go and do this. But now that's an evolution of that where this is the scary piece. Right? Not not even a scary piece is that this is the evolution of this technology where it can now surf the web on its own. So if you wanna book a trip to Paris, it can go to booking.com.
32:40It can type in. It can do these things. But think of that, and then let's let's think of a use case for medical affairs, is that I wanna know what doctor Tom is in the last 10 posts he's had on LinkedIn. I don't need to go and look at his LinkedIn, click, and everything. Yeah. Maybe it is it is you know, the agent mode is a little slower than a human right now, but if you can think of how this is gonna get very fast, it'll go and do that review, click through, see the comments, read the comments, and then report back the insights on on, you know, what are the yeah.
33:11So that is the shift that is happening. More compute leads to more use cases and the cheaper compute. So Alright. So, Nandini, that kind of brings me to, like, where are we going? Like, what's the like, so year ago, we had a conversation. Everything's changed. Now we have a new conversation. So fast forward a year from now, what are we gonna be talking about? A year from now, the it's definitely been, I think, more of a shift in, like, a mental
33:47mindset Okay. When it comes to, like, starting to use AI more in your daily in your daily life. So, you know, you're gonna have people who, you know, who are, like, you know, fully onboard, full adopters of this technology. Then you'll have, you know, the others who are not convinced of it. I think I'm starting to see more people come coming in and then, like, okay. You know, it's here. AI is here to stay. Companies have invested a lot of resources into into this. And then so, certainly, we're building to our AI literacy skills, our skill sets, using these
34:21things. It's becoming a lot more easier for when you're using and interacting with these language models, like Vivek was talking about before, where it's like, you've seen, like I've demoed some of the capabilities and how much it goes into this end up thinking and just doing all the work for you. And then so it's like, the more the people see, like, the value behind it as we're working towards, like, we've gone from use cases. Like, we're moving past that. Now we're looking at building, like, full level, like, enterprise solutions.
34:52I mean, that's where the companies are working towards. Like, how do we make this solutions that's easy to use for the MSLs, where it's not just another tool, but it's a tool that's easy to use, that's gonna be valuable for them. And then so behind the scenes, it's lot of work that's being done where, you know, it's a when you're creating these projects, there's a lot of design thinking that goes into it. And then so that's where we were you know, you were looking at the route of the agents, and these agents are becoming, you know, really a lot more smarter and a lot more and helping I
35:25mean, more it's really simple to build them now, actually, because the information that I was showing you, I demoed, then you can take all that information and then tell the Chat GPT to help me based on all the information I'm working on this project, help me create an agent. And it'll give you the set of instructions to create the agent, and then you can go ahead and then you can do it. And that's you know, that doesn't require we just have to spend time doing it. I don't think it requires a lot of skills to do that. So we're gonna move towards more towards that, where people are, you know, becoming more
36:01comfortable using this technology. Like, I've seen it where all the way down when I go to medical conferences, there is always a talk on AI track and those rooms are full. And you'll see it goes all the way from young residents who are trying to learn it to older physicians, and they're on their phone trying to take notes. Same thing, this AI is being talked about. I was surprised to see there are a lot of AI topics at the Biopharma Conference.
36:31And then so everywhere, every industry is talking about it. So it's definitely here to stay. And I would say that you, you know, if you if you completely shut it out, then you're gonna fall way behind it. It's not a you use it as a thinking partner, as a collaborator, and then help to help you towards working working a different different way of working, a way to give you ideas and to become more efficient at your job.
37:02That's kind of where I've seen the industry trends. Provider health care providers are using CHAD GPT, like, during the ACP visit during visits with patients, and they'll tell me this when I meet with them. They're like, you know, when if they'll like, a patient has a question on the product, they have ChadGPT open in front of their screen, and they're just typing away and asking questions, right, because they're not the expert in, like, every therapy that's out there. Similarly, when they're training new providers, they'll tell them, like, hey, go you know,
37:32instead of the, okay, there we have UpToDate, but then, you know, use TransGPT to help you learn as well and then come back with me, come back to me for more questions. And then so it's being used on the healthcare provider side. So and for for us on the pharma side, then if we have to also learn those skills as well or at least learn the terminology. Awesome. Man, you guys are amazing. Last question. I'm down and I'm curious. So I know there's a lot of you know, we hear ChatGPT.
38:03We hear Gemini, Copilot, Perplexity. There's all these these tools and LLMs. And the and then there's there's apps, right, that are AI focused apps, notetakers. What what would you if you could share some of your favorite tools, like, hey, guys. Try this. Anything that you could share that might be helpful for folks? Well, I'll I'll just share my my current personal favorite, Nano Banana image tool from
38:37from Google. It is an incredible tool for image editing. It is it is sort of a a you know, normal, like, very comfy image editing tool. This is does she talk to it, and it will change the image. Now it is not meant for medical affairs. There's no there there if I wanted to bring, like, a a parallel to medical affairs use case for content generation, right, it would be pretty pretty cool to generate really cool graphics, and you can continue to edit and modify them. If folks have not tried it, try it, upload it.
39:08You know, we can put the hat on a hat on Tom right now. You know, it's it's very easy. Includes the likeness. It's it's just the image editing tool within within Gemini. It's called the the code name for that tool is called Nano Banana, and you'll see a little banana feature on it. So that's one my favorite. But to be serious about, like, it's not the tool. It's the mindset. I think that's what we've talked about. Right? And and then Nandini is just a perfect example of someone who has found time outside of a normal job to learn this language, to interact with this very important part important
39:41technology in human history. Yeah. So that is what's gonna take. Right, Dom? Yeah. What about you, Nandini? Any any any secret tools? The no. No secret tools. I would say to start, like, you're using a WeChat GPT account, start, like, use building custom GPTs and ask it. You don't have to do it on your own. Ask it to help you build a, like, hey, I wanna build a custom GPT to do this specific task, and then it'll help you build it for you.
40:11So the more you practice working, like, building custom g GPTs and that's the same skills that's required to build agents, you know, use that. The I I would say, well, the other tools that I would use out there is Notebook LM, which has become more very common where you can just put in things like, you know, YouTube videos, and you can put in upload documents in there and then have it or publications and have it generate a, you know, very a pod a really fun interactive podcast for you.
40:45And then so that way you can learn on the go. So that's I sometimes I tell my pro like, my my resident like, residents who are learning providers who are new to the field, like, that's one way to learn because we we all it's very difficult to find time to sit down and learn. And then so it's really fun to use that tool. They make it really simple. So go ahead. Like, try out those tools. They're a great they're a great way to start. Lot of when I teach AI is, like, we always start out with fun things and, you know, always, like, look at, you know, look look at doing something do something fun with it first
41:16before you get on to the more serious things. So, like, Vivek was talking about using Nano Banana it's a fun way to start out, and it's a fun way to, like, get inspired by by it. And then you move on to the the more, like, harder, more, like, serious, like, heavy work related stuff. You guys are amazing. Thank you so much. I I'm so grateful that we did this. Nandini, you're a great friend. And, Vivek, always great seeing you. And I'll see you guys in January because I'm I'm actually gonna be at that.
41:47Are you guys both gonna be at that conference in Philly? Yeah. I am co chairing it with Katrina. I don't know if that's been announced yet. But yes. Well, it's starting to it's starting to come into fruition because they asked me if I would participate and help co promote it. So I'm gonna get involved. So you guys are hearing it early. Nice. But it's gonna be in January in Philadelphia AI and medical affairs. It's probably the third or fourth. Fifth. It's the fifth word. Is it the fifth one? Okay. Staying correct there. So the fifth annual AI and medical affairs conference.
42:22We will be there. And, Pacino will be there. So I'm I listen, guys. You gotta get in the right rooms, man, and have the right conversations. So that's gonna be a good room to get in. Once again, appreciate you guys both. Thank you all for listening to the podcast. Thank you for sharing it. I appreciate all your support, all your communication on on LinkedIn, and you guys are the best. Love you all. I'll see you next time.