Sara Rodrigues, Associate Director of Medical Affairs, Vertex, on Becoming an AI Champion.

Meet Sara And Her Role

My name is Sara Rodriguez, and I'm Associate Director in Medical Affairs for Vertex Pharmaceuticals.

Amazing. Sarah, thank you for joining us today.

No, thank you for having me.

Okay, I uh am so fascinated to have this conversation. You and I spoke um, I don't know, maybe about a week ago, um, and and our conversation has been kind of rattling around in my head since then. I have a whole bunch of questions about your AI journey because we often have these kind of enterprise-level AI conversations, you know, head of AI for AstraZeneca or the how do you move the whole ship? We're gonna kind of have like a very human-level conversation because we're gonna talk about you and AI. But before we get there, maybe uh situate us in um, other than the AI part of what you're doing these days, what what what do you do at Vertex? What is your job? Uh, where are you in the world? Give us a give us a sense of uh of Sarah Rodriguez as we get started.

Thank you. Happy to do that. I think it makes total sense to just contextualize everything. Uh so I'm I was born and raised in Portugal, just to set it up in terms of moreness. And I studied there in Portugal as well. That's where I did my pharmacy degree. Uh, but eventually my brother was already living in London and he kept you know poking me. You should come and join me. Yeah. And so, you know what, like why not? Um let's stop living the dream here at the beach and uh actually embrace ourselves into a uh a proper a proper career. Um this was back like 10 years ago, so uh things have changed anyway anyway. But uh off I went to to the UK. I I I worked in pharmacy, uh retail pharmacy, but also auto patients in in hospital settings. Um, and that was very cool. You know, the system was very different uh if you compare like UK to Portugal, so that was a really huge learning curve. But then uh, you know, pharma, pharma came came to me, or I came to Pharma, I was quite excited to try new things, uh, you know, uh just before you hit your 30s, let's try something new. And that's when I started it at Vertex. So that was about seven years ago now. Uh, and it's been a fascinating journey working at Vertex. Uh it's uh, I suppose at the time was a bit more smaller biotech. Now it's uh really, really growing uh in a fast, a fast pace. And um now I focus mainly in gene therapies and based in the international head office. So that's that's really cool, you know. Pretty much everywhere apart from the US. So that's that's super exciting.

Why She Spoke At Age Of AI

Um what I mean the number of of people that we have on the on the show that that have started either as pharmacists or in retail pharmacy, uh, and then are like, I think I'm gonna I think I'm gonna head on over to to pharma. Uh very interesting. So um, okay. So we're we're talking to you today for a bunch of reasons, but but primarily because you uh recently spoke at the age of AI event in London um uh and got fantastic feedback on your time on the stage. Maybe for those of us who weren't at the age of AI in London, what what were you talking about? What brought you to that event? Because it wasn't necessarily your associate director of medical affairs position, although that's a bit of it. Um what were you talking about and and um yeah, like get us get us to there?

Yeah, no, so I think age of AI was really exciting and you know, AI, everyone is talking about AI these days, but I I was quite lucky to be part of um you know a pilot scheme uh at Vertex and starting using um co-pilots and experimenting, and I think I was always quite uh you know excited to learn about AI and new technologies. Uh a couple of years ago did a no-code uh course as well, trying no code tools to build like websites and things like that. So I've always been super fascinating with technology. Uh, I suppose part of it is my brother's influence because he's a computer scientist and uh solutions architect.

So that'll do it. That helps.

That helps, yes. We do bounce lots of ideas about AI, but it's quite interesting to see how much sometimes I know that he hasn't picked up yet just because it's such a fast pace, and I've just been so involved in it from a different perspective. Uh, but going back to age of AI events, um, I spoke about using AI at the individual level, so at the individual contribution, and I think that was probably the big difference as in contrast to you know some of other talks which talked about you know bigger solutions, which probably my brother would be more familiarized with. Um and so I was talking about how one can create agents to support their day-to-day role, you know, be it a brainstorming agent, you know, using AI to brainstorm with you ideas, to refine ideas, or an agent that pulls all the information about your product, for example, so that you don't have to go find the information elsewhere or flick through thousands of pages, uh, or just an agent on how you can uh summarize everything or all your thoughts and notes from a Congress session, for example, and quickly turn that into five succinct bullet points. So that's essentially in a nutshell what the talk was about. Uh almost like having three to four AI assistants working for you or with you to just really become more efficient in your day-to-day role. So you're right, not just exactly my role as uh A D uh in medical affairs, but also how you can, you know, anyone really in medical affairs can leverage AI to support in being more efficient.

Pharma Guardrails And Experimentation

Okay, so my my million questions start start now. So um okay, I I wanted, I'm gonna uh I'm gonna sort of frame this up kind of in in in two ways. So um first is you know, I have conversations like this a lot, and it feels like uh generally um kind of people fall into um like maybe three different categories, and I'm really making this up on the spot in terms of their AI usage, right? There's the there's the like either sort of no or or very light usage, right? So um haven't really done a lot with AI or like use ChatGPT the way I used to use Google search, right? You know, there's that category. Um and and there's a lot of folks in that category that hear things like vibe coding, and I've got an agent, you know, doing something for me while I do something else. I think that in for for those folks, that whole world seems incredibly foreign. So I want to talk a little bit about how you started to get there. But but then there's this other category where I think people have gotten on the other side of that and they're starting to uh experiment. Maybe they're you know, they're they're vibe coding something or they've they're they're sort of training an uh you know an agent or um uh you know a task-based uh uh you know kind of application in AI. And then there's a very, very, very small, I think, third category of people who are like really, really dialed in. But that the population of folks that I talk to feels still, especially in pharma, very skewed to that first. I kind of don't really know how this stuff is gonna help me outwork category. So um, so one, I I want to maybe talk about how people kind of make that leap. But before we get there, this is one of those questions that Daryl will edit down because it's so long. Um uh before we get there, I I I do want to like ask you about Vertex's comfort level to the extent that you're that you can talk about it, with you in your role uh leveraging AI. Because I still think there's a lot of organizations who, you know, sure they've they've said you, you know, you can use Copilot. This is our corporate kind of there's a check mark beside that, but only in certain areas and only to a certain extent. So it sounds like you're becoming a bit of an AI super user within a pharma company. Can we start with the pharma company? What what has the what has the comfort level been like within Vertex? And has that kind of changed over the last little while, uh, opening up more opportunities for you to experiment?

Oh my god, I'm not sure I can talk too much about that. That's fair. That's fair. Uh I think what I'll I'll what I probably can say, and let's just revisit to to see whether let's just revisit it after I finish. But I think uh perhaps overall as an organization we were quite uh conscientious about the um conservative about the use of AI. However, things have moved, have moved, and as have changed you know very rapidly, and I think that's probably very similar across many other organizations. There are rules in place, and I think as long as employees follow those rules, employees, employees are happy to experiment and um put forward their ideas to to trial them out, let's put it that way.

First Steps Beyond AI Search

I think that's a great answer. So in an environment, so then I'm gonna go back to the first part of my long question. So in an environment where um there are more opportunities to utilize AI within a very you know clear rule set, um where where you know where would you suggest or or in your experience, where are the most impactful kind of first steps to be found as you um move beyond kind of using AI like search and moving AI into being actually like a work partner?

Yeah, I think as much as I'd like to say people are free to experiment and do their things as long as it's they stick to their SOB. I think the problem is that the visualization or the understanding of what AI can do to help at the individual level, people in their own roles, it's limited just because of not understanding the possibilities of it. I think that is the problem, is where do you start uh beyond utilizing it as a Google search, as you say. And I think that's the key shift that people are not yet doing. Uh, and this is where you know, perhaps IT departments, from my understanding, are you know somewhat disconnected from you know the subject matter experts or the people working in medical affairs or the people working in the marketing or regulatory, and this is something that I hear across you know all organizations, so it's really important to like connect those dots, sharing, creating uh AI champions community and things like that. But I do think, as you said, the majority of the people are still in that first bucket and are slowly transitioning to the second one, yeah. And to be honest, for for me it was I think part of it because of um I was uh lucky for like two years ago to just realize I was maybe being left behind. I thought, gosh, like, am I this old that I don't know about all these tools that are out there in the market? I used to be so savvy when I was younger, and so I was like, gosh, I need to have skills. So maybe like that was the reason why I got excited about it, just because I took took that um, you know, extra on my personal time. I took took that extra time to do this no codes tool course back in 2024, which already, as you may, as you may know, like many of the tools out there already had like the AI chatbot. It was just as refined as it is now. You couldn't just vibe code a website. I mean, you could you could try, but it wouldn't lend you anywhere. Um, so I think that was a a bit of um being lucky to have done that and understand the possibilities, and then just like started seeing and learning, connecting with people that actually were experts, uh, just in general, not specific to pharma, and then connecting the dots. Oh, it can do this for marketing, it can do this for sales. Oh right. So let me have a look at to see what it can do for me in my job.

And could you give an early example of okay? I've done this uh, you know, this no-code training. And it's it's two years ago in the world of AI, seems like we're talking about, you know, 1974, right? Like it's just I I think have had changed so much and continue to change so much, but but you know, you've done this no code course, uh, you're starting to think about, okay, how can I apply this um, you know, in in my sort of either personal or professional life? What were some of the early um like magic moments where you're like, oh wow, like this is really impactful? Do you do you remember moments of um experimentation where you really like the kind of lights went on?

No Code Automation Aha Moments

Oh, totally. Automation was one of them. What like I can just automatic automate an email out to someone once once the trigger is actioned. I did I, you know, like these little things that may seem so basic for someone working in IT DTE, but if you if you're not aware of it in medical affairs, which is something that almost feels like a dinosaur in a company, uh, I think these things do create space for innovation in your in my at least in my mind did. So just by simply creating um at the time I I created a simple platform to track your prescription. So from the moment you sent the prescription to a pharmacy to the moment where you could collect collect it. This is something that already exists in the UK, but it didn't in Portugal, so it just made sense to create something like that. Uh and I and I was you know fascinated because how can I, pharmacist understanding nothing about websites, could create this? Um, and that was for me a waha moment. And automating uh, you know, the emails, it's ready for you to collect and things like that.

And so you were doing this like for the pharmacy that you were working with?

No, no, no. I wasn't working in a pharmacy, it was just an idea to experiment uh and put a product out there. So no, I was already working at vertex at the time.

Okay, okay. But I think that so I think that the the there's so many important things in the sentence that you just said, but one of I think the most important things and again where I think people get a bit stuck is the, you know, you said, and you know, I'm paraphrasing, how can I, someone who doesn't know about building web, how to build websites, how could I build this thing? Like that's when the lights went on. Um that that is uh like to me, that is the the promise of AI, right? Is that is that you don't need a deep technical understanding to be able to do uh are really quite technical things because the AI will do the technical part for you. You know, what you need to understand is that there's a there's a workflow problem or a business problem or an information gap, right? Um and it's that understanding, it's the deep understanding of a s of a system or a process that the AI will help you solve. And you know, I think anybody who's in a you know in a pharma organization who's who's you know, they don't even need to be very senior, if you tap them on the shoulders and said, uh tap them on the shoulder and said, you know, is there is there a problem that you that you think needs to be solved here? Is there something in your day that could be done more efficient uh more efficiently? You know, they would give you a huge long list. So how would you suggest imagining that somebody works in a far company that has a pharma company that has no AI rules, you can do whatever you want with whatever tool you want. How does someone who's never um uh you know taken the steps that you've taken, who also doesn't know how to build a website, but who also has an idea where would you suggest they start to experiment so that they can have that aha moment?

I think aha moments need to come from seeing examples. Okay, you know, like understanding what it can do for you. But if you see what it can do for you, so you know, automate uh doing an automation or a task for you is possible, then evaluating your own workflows, what is repetitive, what is like burdensome in terms of admin. I don't know, let's say uh Slideck is approved. Well, can an AML be automatic automated to the whole team, just letting them know instead of you writing it, for example? I'm sure it can. Uh, you know, it's maybe sometimes it's not a a huge thing, but tiny things that incrementally will save you that amount of time. But it all comes from understanding what it is possible, and you know, like automation doesn't necessarily mean AI, it can just be an automation, but uh uh it's the understanding of what is possible that sometimes is missing with within you know with for us, it's it's missing, it's that connection.

Building Trust And Protected Time

Yeah. So as a as an AI champion, I feel like you you you have you have like a big medal or like a hat that says AI champion um in vertex. What what what does that mean and and what what how do you interact with the rest of the organization to help folks um you know have those moments of enlightenment themselves?

Yeah, that's uh um you know I would I don't want to say million-dollar question because I would like to think that I have a plan. Fair enough. So I think it's a a lot about what we've just been discussing, you know, showcasing use cases across uh the different functions to sort of bring that those waha moments, but also there's a a a huge important gap in terms of you know those individuals that are indeed worried and they're almost like, I don't want to look into this because to be honest, uh one, I'm scared, will I lose my jobs? Will it replace me? So it's it's really increasing trust as well for those for those colleagues. So it's not just about sharing, oh, look, how brilliant um this is you know doing things for me, but also, you know what, you this can help you. It's not going to replace your job, but it can save you time and you can be more efficient, you can have more time for things that matter. So it's shifting as well that perception, otherwise, people will not experiment. And the other point, and actually, we talked, I talked about this yesterday, um, is the fact that it does take time as well. It takes time out of your role to experiment, to fail, to tweak, and make it better. And and if you just try it once, you'll be like, Well, I just wasted the whole afternoon and I landed on nothing. So it's really that I I hear that some companies are like um providing some free time in brackets, so like providing uh um protected time, for example, for AI enablers or AI champions to experiment. And I think that's probably super important as well, uh, because it is an effort. So I think it's those three things is the making sure you create a safe space for those who are more scared of it and you know, bring HR into conversations, be open about it, be transparent what the company vision is about it, make sure that people understand that, you know, hopefully it's not here to replace them. I think in pharma we have a long way to go because we it's a heavily regulated environment. So I don't think humans in the loop will go anywhere anytime soon. Uh, but but I mean, uh who knows because Claude keeps reinventing itself. Uh and then the second is really uh sharing those examples to sparkle, sparkle those waha moments. So this is possible, and you don't have to be a tech person. Um and uh making sure people understand to to evaluate their own workflows, what what can improve, what can help, what is good, what is not good, and making sure they they use it safely as well. Is it super clear what is what you should and should not do with it? Um just so it's comfortable as well to experiment and not doubting yourself that you can indeed. Upload this document, but maybe you should upload that document and you should only use the tools that are available at work. And last but not least, protecting the time for those that are experimenting because it's not going to be efficient from you know day one.

Experimentation Habits That Actually Stick

It it's interesting. Like I there's a I think there's this really um sort of symbiotic relationship between a couple of the things that you just mentioned. So the sort of like use cases, experimentation, and kind of like potentially wasted time going down a bit of a rabbit hole spending the afternoon and not not getting the right thing. And and uh you know, the way that I've been um sort of talking with with with folks in my organization as well as as you know other people that I interact with is is is that um you know I've come to look at um how you experiment with AI as um a bit of a a way of thinking versus um a set of use cases that may or may not directly apply to you and how you know this is someone used it this way and could I use it that exact same way? And it's so it's more like the use cases to me when you talk to folks about them are more uh illustrative illustrative of I was trying to do uh something more efficiently or I or I recognize this opportunity, I then sort of did this with AI, and it resulted in the following um uh you know efficiencies or or um you know outputs or or whatever. Because I think that sometimes where people get stuck is that they're taking the use cases as as as you know very, very discrete examples, like this is sort of quote unquote the thing that that AI does, and not taking those use cases as a way that someone has applied um uh a new way of thinking about solving their problems. So we have a like we have a 30-minute scrum every week where basically people um we all get together and we talk about what is something that everybody tried differently this week uh using AI. Did it work? Didn't it work? And so it's more about building that the right kind of experimentation so there's not wasted time.

I love that. I shall I shall take that example with me.

You're welcome to take it. And I like one of the I mean, one of the examples is is the thing that I I would say to my team is like every Friday, you know, one, you you can use AI to tell you where you can use AI, right? So I was you know, I was saying take a screenshot of your of your calendar and every Friday uploaded uh into you know Chat GPT and say, this is what I did last week. Uh, can you suggest five areas where I could use AI to be more efficient, right? Like there's there's there's ways that that you don't need to be the one to come up with like a new way of doing your job. Um, but you know, we I gave someone the example the other day that we had to do an edit of a whole bunch of these podcasts. Uh, and you know, frankly, it was a little overwhelming to think about listening to myself for an entire Saturday afternoon. And so I just took the transcripts that that our podcast platform produces, and I uploaded like 25 transcripts at a chat GPT, and I said, pick you know 15 of the most insightful comments, uh, you know, give me a timestamp and and an episode number. And I poured myself a cup of coffee and it was it was done. It wasn't perfect, but it saved me hours and hours and hours. And um that was just born out of a, oh, like I don't really want to do this. Is there a better way to do it? And I think that if you're supporting folks in spending more time in the oh, like what is it you don't want to do? What is what is it that could be made more efficient? That's almost a fantastic starting point for how to get some AI adoption.

Oh, I love I love that example. It really reminds me of uh what Stephen Bartlett's team was doing for them, for him, like very recently, because I think they did something very similar. Uh no, but I I completely agree with you. Uh I think finding those those tasks that you don't really want to do, it's it's a very good starting point and also a very good selling point just for those that do not want to start using AI. Uh, but also to the other point that you're saying, you don't have to be super like you know prescriptive. Oh, this is the template you should use, uh, allowing people to understand that you know you can just ask it how to do something. Uh someone was asking me uh a new joiner in in our team uh in terms of tagging things in Viva, what good looks like, and I was just like, you know, look at you can upload the SOP to our internal tool and you know ask it the best approach uh to do it. And it was like, oh, you can do that with you know AI. It's like, yeah, yeah, yeah. Um, you know, it's just a tiny thing that sometimes it's just understanding that you don't have to know what you don't know. You can ask the AI to help you with with those exact things, or just to make something more efficient, or you have a uh a problem you don't know how to solve, you can ask it how to solve. But it's just knowing beyond and seeing beyond the Google search that is the problem.

AI In Real Life Logistics

So so you you write a Substack on AI, if I'm not mistaken. Fear not okay. And I also saw a recent LinkedIn post where you were sort of saying, um, outside of my work as a mom, like Claude was really helpful lately. Can you give a couple of examples of where it still feels like you're having these moments of, oh wow, like you gotta AI sort of swooped in and saved me a little bit there. Um, because I I think you probably have some recent examples as well.

Yes, that's a very good one. So my son turns three tomorrow, and we have a birthday party planned for Saturday. Uh, but I I mean uh there's a lot of logistics behind the birthday party, but this is my first birthday, proper big kids' birthday party, right?

So there that's a lot of pressure for a birthday, I'll never remember. But yes, yes, I know.

Uh so I started getting scared because all these parties we've been, there's lots of animation, we had nothing booked, just a place we we rented. Um, and I've recently visited the place, and it was quite, you know, it was bigger than I thought was bigger than I thought. Second, uh, they kept asking me about animation. I was like, oh, I don't have anything planned. And secondly, and then thirdly, uh, you can send us a sketch of what you want uh how you want the tables and the chairs, everything laid out in the room, and we'll do that for you. And it's like, what? I mean, it's in three, it's in three days. I've I haven't had the time. I have a full-time job. What do you mean? So my immediate thought was uh I spoke, I say speak to Claude because I I I couldn't really, you know, as a mom, you're either like carrying uh a kid or you know, pushing your buggy. So just voiced to Claude, this is my problem. I have this room. My son is turning three, as we have these birthday parties, so many kids, um, these are the numbers. This is what I have, this is what I don't have. Uh, can you help me? Can you do a sketch of the room? What are your thoughts? Things to do, things to organize. And yeah, I must admit, like, I was it's just one of those things that you never realize that you're gonna get a good output until you see the output, and it was great. A sketch, perfect. I mean, I uploaded a photo of the room as well, so uh yeah, just gave me the sketch. I just forward the sketch to to the lady with the tables. I'm gonna do like I bought two small football nets to put in inside the room. My partner is gonna be the football uh coach. Um, yeah, so we have like a whole like um party um schedule, and you know it's great. Um, and then next next thing I did was create me a grocery uh shopping list. And I mean, either I'll give it to my partner, I'll give it to Comet to put it onto our shopping online shopping list. And yeah, that's that's it. Done sorted. Let's see how it goes. But see, I so I think that that's so that's a perfect example.

Prompt Hype And Free Learning

So I was, I was, I was taking notes as you were talking, because it look, there's gonna be a bunch of listeners who are are immediately after this gonna be like, I'm gonna be like Sarah, and Claude's gonna completely help me with my next kid's birthday party. Um, there's other people who uh you know don't have kids or their kids are are planning their own birthday parties because they're they're they're teenagers. But but if you break that down, you know, the the fundamental elements of that feel like the um like the ingredients of how to experiment with AI, right? So you were you were stuck with um an ask that you were actually like quite capable of doing if you had the time, right? You you seem like an incredibly capable person. I'm sure you could figure out where the where the tables went in this room, but you you didn't have the time, and and probably of all of the things that you wanted to do, like it was not super high on your list. So, you know, here's a here's a here's a a problem that you uh likely know the kind of steps to solve, but you don't really have the time. I think what's really important, one of the things that was really important that you said is that I I I talked to Claude, right? So I think that that folks who aren't um uh uh you know sort of AI literate or aren't using it often, uh, you know, they get caught up in this, like, oh, there was all this, you know, prompt engineering and it's all about my prompts. And it it's just it's actually you can just turn on voice and like talk, right? So that I think is a huge, uh a huge part of it. You also just explain the problem, right? You you it wasn't a really structured, I've typed up a document with all these things. It's basically I I started talking to Claude with my voice, I explained this problem, I explained what I was hoping to get out of it, and I, you know, I uploaded one picture, right? Like I gave it one asset and and it it it essentially maybe there was one tweak, but gave you very quickly the results that you wanted. And I think that that that sort of breaks down a couple of what I uh hear often as the like I don't know where to start because it's the I don't really even know how to interface properly with these systems, right? So you can just talk to them. And I don't know how to get them to do the things I want because I think that there's still this sense of, and yes, prompts are important, and like when you get into stuff that's a that's I think a bit more higher fidelity, but the the early kicks at the can like can literally just be a I'm talking about a problem that I need solved, right? I think that's that's really important. Do you find that um people overcomplicate when they're when they're new to using AI? People uh make it feel more complicated than it needs to be.

I I think you hit a very important point. I think there's a whole industry now being built about around how important it is to do XYZ with AI tools. Do you see what I mean? Like exactly what you were saying. Do we actually need the need a one-page prompt to solve a problem? Maybe not. Maybe it will be helpful for certain situations, but I think actually, as we can see, AI is evolving in such a way that actually, at least with Claude, I don't feel the need to give it a uh a four-page full of instructions. Um, at least not from for what I've been using it, and and mostly it's been for my my personal my personal things. I don't use it for work. Uh, but I actually think these tools, and by the way, Copilot now has uh you know agreements with anthropic and and claude models are incorporated into Copilot. So it's not like we're talking about such different tools, these tools are becoming more inter interlinked and intertwined. So I do think that also that industry that you mean you open LinkedIn and there's so many, oh, there's this A4-page prompt that's gonna you know create a business for you or do this for you, and it feels like sometimes is unachievable if you're like starting from zero, starting if you want to start it, people don't know where to start. And this is a very common question: how do I start? Do you can you recommend uh a course? And and to be honest, open AI as free courses, yeah, topic has free courses, it's just people don't know about it, and um, there's a lot of people selling it. And the other point is you pay for something and you go to that, you go pay a webinar, let's say, and this is very common these days, you pay for a webinar, you go to that webinar, and at the end you're say it someone is selling you another 12-week course after that. So it's like, well, where does this end? Is there like uh at the end that is like a light at the end of the tunnel? Do I actually know everything? But to be honest, I don't think anyone currently knows everything about AI because it's just literally impossible.

Well, and it's also uh like there's a there's a um there's a threshold, I think, that is um actually I I think quite easily met in a lot of cases that will get you uh kind of almost everything you need for the time being for AI adoption in most roles, right? Like there's a there's a it it it it it it doesn't feel like there's a really big hill to climb to start to actually be doing what you did with your son's birthday party, right? Which is which is I have a bit of a go-to tool, I interact with it in a certain way, I am, I'm I'm building habits around asking that tool to help me solve a problem, and I'm doing it in a way that is not phenomenally structured, and that's delivering like pretty good results, frankly. And I think if more people got to there, right, then the incremental gains uh will will will I think just be natural, right? Like it's a it's a bit of a um, you know, the analogy that I gave to somebody the other day is is uh um and I don't know if you take transit or you drive to work or you work from home, but but um, you know, I I have an office that's close by to my house. Uh I still drive, I'm too lazy, my apologies. And um, you know, there's days that frankly, like I sort of I get home and I haven't really like, did I even like I just I'm just on autopilot, right? Um and I think that um there are uh tasks that we all do in our jobs that we're that we're sort of on autopilot for, right? It's just this is the way we do it. And I think a big part of AI adoption isn't a technical skill set, it's a curiosity to s to start to look at the things we do and say, why am I doing it that way? Right? That that feels honestly like the biggest shift. I don't think it's about taking a first online course, and I don't think it's about about prompt engineering. I actually think it's about I'm stuck with this stupid layout for my kids' party, and I'm not gonna, I'm literally not gonna pull up a pen and paper. I'm gonna do it differently.

Accuracy Risks And Humans In Loop

Yeah, I completely agree. And to be honest, that these tools are getting so much better as day, days, months go, that maybe someone that was frustrated like three months or six months ago when they tried AI for the first time, uh, eventually they'll go back to it with another problem and actually will retrieve really great results, and that person will be bought into trying it again. And yeah, I think maybe people struggle now, but eventually it will become so much easier and AI will become so much better that it'll just be a super straightforward, you know, go for it. And uh maybe at that point we don't even need curiosity because it just will just be so automated. Um yeah, I think I suppose the only problem that's that's probably here to stay it's you know, hallucinations and just making sure that you know, I mean, I work in medical, so as you can imagine, it's quite um important thing to get right, accuracy.

Yep, yep.

Uh, and I think knowing that is, you know, probably the most important bit, and you know, just really avoiding um decre uh how do you say that? Um we're probably gonna cut here, but is making sure that how a function is viewed doesn't uh how a function is relied upon doesn't get you know uh doesn't doesn't decrease or it doesn't AI doesn't impact it. Do you see what I mean? How someone trusts a function doesn't uh decrease just because people are using AI.

For sure. Yep. And I I mean one of the ways that that that we're thinking about it is uh and again, I mean like we are very early in in an in an AI journey. Although I will I will say that um I got really super sick of hearing about people vibe coding. Uh and so like a month ago, I just whatever, I bought some credits on Lovable.

Um and I should should I've tried the Women's Day. It was for free. That's what I did.

I don't think they let me into Women's Day. Um uh and I like I bought some credits on Lovable and I just I I just built something. And it was like that was my magic moment. I was like, this is nuts. And so now my whole team is is building applications and they're vibe coding stuff, and it's just it's like we have turned it, turned an unbelievable corner, and partly because it's it's we're just in the like we're gonna do things and stop talking about them. But the thing that we're not doing, and I think this this draws a straight line back to what you just said. The thing we're not doing is we're not uh using AI to create the actual outputs, right? We are still having uh, you know, uh human talent and human eyes and human judgment to create the actual outputs. What we're using AI for is to make a whole bunch of efficiency gains in the process, right? That that is um is is that surrounds an output. Um and that that will likely change, and maybe that's not the right thing to do. But we've sort of been like, okay, how do we create AI efficiencies in the briefing process so that we can be more efficient getting to the creative process? We'll still still do the creative process in the time being with humans, but there's lots of efficiencies to be gained. And I think that that's maybe a bit analogous to the like you're in medical, there's a bunch of things that that that you still need to do as people in that uh craft.

Oh, I completely agree. But I think you know, pharma is uh I would say building apps within your work environment, it's quite restrictive. Yep. Uh, but I mean I mean 100%. I would I've I would love to build an app just for myself to organize things through uh and not having to you know connect to multiple systems, for example. I think that's probably uh something that every organization needs to look at. But yeah, I think we are entering a very interesting moment, and I I love that you allow you know everyone in your company to you know vibe code their solutions. I think that's that's brilliant. But I completely agree. We still need to be those you know, creative minds or leveraging AI as a creative partner and still building those outputs because otherwise, and and I I think it was recently I saw uh someone posted some results from Anthropic and how nearly 80% of a medical writer's job could be replaced by AI. And I'm wondering if the human element is just really the sign-off at the end, how will these people maintain engagement, curiosity, uh, and you know, in just enjoying their role, I think will be really difficult. So is reman reman remanaging, oh uh sorry, revamp uh just revamping work. Job roles will be probably super important to be to keep people engaged.

Yeah, and and and look, I mean, just because you maybe can doesn't mean you definitely should, right? So sure 100% of a medical writer might be able to be, or whatever, 50% of a medical writer. Well, maybe there's a bunch of other gains to be made before you start to automate all of your medical writing, right? And and and again, there's there's there's I'm sure there's people listening thinking um, you know, this is way old school thinking, and you should automate everything. And I just I think that especially being in life sciences, uh a big dose of uh caution is is definitely uh is definitely required. Um the the other the other thing that I will I'm gonna I'm gonna share a thought and then and then we'll sort of end maybe on going back to to your experience with with folks in your organization. But uh, you know, you mentioned that it's that it's um it's something that we're that we're encouraging everybody to vibe code. A light went on for me the other day as a as a as a business owner and as a you know as a as a manager, you know, one of one of the things that I've always strive to do in the businesses that I've that I've owned is um you know ensure that that folks who come out the other side of their time uh in my organization are um you know infinitely more valuable in the marketplace than they were when they started with us. Um, you know, my I think part of my job as a as a business owner is to um have people feel like their career advancement and skills development while spending time with with uh with our organization is tremendous and almost tremendous to the point where like they don't want to leave, right? That should be one of the areas of retention. And the light went on for me the other day that I think I would be doing the team a disservice if I wasn't creating an environment where they could become very AI literate, because if they're not very AI literate, I think that their uh employability will be compromised when they're on the other side of working for me. So it almost is it's it's got a yes, like super important for the business. I want to be able to turn to my client and show them efficiencies. Like there's a there's a whole bunch of reasons, but actually one of the reasons is I think I think employers have a responsibility to their employees to have them skilled up and ready for what is the biggest transformational shift that you know most of us will ever see in our lifetime. Um, and so that's that's a part of it, is is I feel that responsibility. So anyway, that's a bit of a tangent, but it will are all tangents, it's okay. Um, but it brings me back. So it'll bring me back to could you share an example or two again? Like, don't need to name names and and but like could you share an example or two of where you've seen folks go from really not being comfortable using sort of AI in their daily workflow to having moments of of kind of like you had with your son's birthday of like, wow, this was really important. I want to leave listeners with maybe a couple of those use cases that you mentioned. That isn't like, oh, I built an agent that's doing my work for me, but it's like, oh, this was a really helpful first step. Like, what are some really helpful first steps that you've seen folks take within a pharma company?

Simple Pharma Use Cases To Copy

Um I can I can tell you about uh a webinar that we did in January, uh actually with Farber Brands, where I also talked about the product expert uh agent, which isn't, you know, it's not an agent that is gonna do automations for you or do tasks for you. It's more like where you can find information about the product because it includes in its knowledge base all the relevant information about that product. And it was interesting because a couple of uh colleagues that I know of connected to listen to the to the to the webinar. And at the end, I got messages from, I'll tell you, like a clinical research associate, so a CRA, was saying, Oh my goodness, it this would be so helpful for me because I'm I keep like opening so many documents to find uh something very specific that this would just like save me so much time, or even just including goldie SOPs or clinical trial regulatory regulatory documents that sometimes are so dense and long. That was one, or just another friend, a pharmacist doesn't work in pharma but works in a hospital, and she's just doing a course uh studying. She's always studying. Uh, but she was like, Oh, I can upload all the slides from you know these subjects and then talk to it and like it would it would ask me questions, or I would ask questions about the particular subject, and I'll just learn learn. It's like, yes, yes, you can do that. So, you know, it was a specific use case, but people could see it in different uh ways. So, you know, just another just a very easy example. Anyone can do it in co-pilot, it's a very simple way to create um a co-pilot agent. It's it's agent, but really could be a custom GPT. Uh and um yeah, I think that's just uh it's a small thing that really could uh could help you out.

Yeah, so so I I think that that you know there's a lot of folks that listen to the show who are on in in brand management positions and and um you know the ability to create maybe if there's somebody new coming onto their team, you know, to create a like a brand agent that has uh you know research, data, you know, demographics, you know, competitive information, so that can all be queried essentially like a chat or a conversational query that's just a a real a very specific brand expert. Feels feels like something that that uh would have like incredibly uh uh high value and and widespread usage.

Yeah, definitely. Or just uploading an FAQ into your chat and ask uh ask AI to ask you questions about it, or formulate uh uh you know uh uh um sorry, artificial conversation between an HCP and yourself, for example, and you can train yourself on those FAQs and how to respond or not and ask it to even score you. So these are just like simple things that everyone could do, even just to upskill themselves.

So I'm going to guess that your Substack has a whole bunch of examples. Where can people find more from you after they finish listening to this episode and they want to get uh some more of your insights?

Oh, thanks for asking. So they can follow me on LinkedIn. I post regularly about some things I've you know noticed, or just a small intro to a Substack that I post weekly on Mondays. So on LinkedIn you'll find me, my name, Sarah Rodriguez, and then you'll find as well a link to my Substack. That's probably the easy way.

Amazing, Sarah. I so appreciate you taking the time, especially with a big birthday party coming up. Um I really enjoyed this conversation, and uh again, appreciate you taking the time out.

Thank you so much, Neil.