WEBINAR: Augmenting the Customer Experience with AI Powered Analytics

Hello and welcome everyone! Good morning, goodafternoon, depending from where we are connecting to this very exciting NEXT Webinar! Verypleased to welcome today actually four instead of five extraordinary pharmaspeakers So from left to the right, pleasure to welcome Sudhir Mahakali fromSandoz Hi, Sudhir, how are you? – Im doing well Really happy and excitedto be on this panel! Thank you! Thank you very much Then were pleased towelcome Alexey Cherchago from Sandoz as well, colleaguer from Sudhir, obviously Hi Alexey,how are you? – Im good Dario! Happy to be here today! Thank you, Alexey.

And Philippe Kirby from MSD Philippe, same question for you, a tough one right? Yes, yes, sorry you cant see me everyone Hello,all is fine! – Thank you and last, but not least, Orchid from ODAIA Orchid, same question for you,how are you and hows weather in beautiful Canada? Its fantastic and I think its actually warmerthan Europe and where Philippe may be.

So pretty excited about that, its rare for us! Great tobe here and very excited to be part of this discussion! – Well, now we are jealous, but I thinkthat some parts of Europe are also enjoying some nice weather Okay, so todays theme: Augmenting the Customer Experience with AI-Powered Analytics I think this will be a quite interestingdiscussion and a big challenge to satisfy the needs and the expectations from the audience andfirst of all, I would like to start with a very critical question which is: COVID generateda huge amount of data, which we all know,.

but how and when AI can benefit from that hugeamount of data? Lets start with you, Orchid So I mean, its amazing how much data there isand it depends on what perspective of data we are referring to So from a position of people whoare experiencing COVID, sadly there’s too much data and I think almost every pharma company, andI cant speak for any right now, because I havent been part of the pharmaworld for a year and a half,.

I think every pharma company is trying tofigure out how to help people with COVID with not just the data, but usingthe data to understand patterns of peoples behavior Not just how theyre sufferingfrom COVID, but also what theyre doing in terms of using resources So how our patients,for instance, are doing very different things during COVID in terms of virtual healthversus going to see their doctors live so that has completely, I think, transformed the waypharma works in terms of geographic, one-on-one,.

face-to-face kinds of discussions So it bringsup opportunities, but I think, from my perspective, Im very passionate about the augmentingthe experience of the customer, being the internal customer as well, so the actual customer beinga sales rep, for instance The world of the rep has completely changed, so I think usingthat data to see how we can actually help.

the people on the ground, the front-facing people,do better during this time is something that I feel that we could do better on ourside, on the analytics side, as well as pharma Thank you, Orchid Sudhir, same question foryou? – Yeah, so Dario, I think you know I completely echo what Orchid mentionedabout the data deluge, you know with respect to you know during COVID times, simplybecause a lot of the interactions went digital right and so you know, you have moreinsights coming in that need to be trapped Now, with regards your second part of yourquestion, on how and when it is available to AI,.

I would say that you know we have this challengethat when we are incubating, prototyping models, I think you know we are able to do well withaccessing the data anyway formal fashion, its available But then when you get into seriousindustrialization more and you want to go live with streaming data or with you know datathat needs to be a little bit more modern for the machines to pick it up, I think youknow we are a good 6 to 12 months away, so.

typically, this is the conversation that we arehaving, that the infrastructure, right, aspects really came front and center now and you knowwithout the infrastructure capabilities on the connectivity front, which is really going cloud,you know that’s the only way there is, to scale And then on the modeling aspects right, to beable to manage you know the graphical nature of interactions between the data, you’re not going toyou know leverage it in an industrialized setting So I would say we are you know, just from ourperspective right, the work needed to get it from.

the current state to a target state is 6 to 12months away – Thank you, Sudhir Lets move from Sandoz to MSD, and Philippe lets hear your point of viewon this interesting discussion so far Yeah, very interesting points of view alreadyshared on the data legend and certainly, I think, even if we had made headwaysin terms of engaging people, data scientists, because I think we need to look at this froma people process and technology perspective, and started getting you know these datascientists to be able to talk to our marketers.

and vice versa and really understand what businessproblems we were trying to solve, we just werent ready for it And so here, you know all of a suddenyes, we were confronted with all this data, all this insights and the need to really exploit it,because all of a sudden digital was really the only means to communicate with customers Andso, you know were kind of caught off guard and I think that’s really, to add to whats beensaid so far, is really that next thing, is to say.

how do we really define what we want to dowith this data? What are we going to use it? So you know the title of this is aroundimproving customer experience, but still, you know more precisely, exactly what does that mean? Whatare we going to do in terms of improving customer engagement? So I think that’s really the next focusand certainly one of the things were looking at.

and you know were going to talk about thingslike next best engagement and so on But thinking through about how we actually exploit thatdata to say how do we add value to customers? How do we give them what they value? Because whatthey value is whats going to make them want to engage with us more and have that bettercustomer experience Thank you, Philippe Alexey? A lot of good points.

I would say there are severaluse cases right related to COVID where AI-based tools and technologies are being used And firstof all, its detection and diagnosis of COVID cases right? And here several AI systems havebeen designed, developed for diagnostic of COVID, using, I dont know, medical imaging technologies,such as just computer tomography, for example, or X-ray images There are also some systems designed around epidemiological predictions, you know predicting, I dont know,mortality rate for example or long-term patient.

hospitalization, all these elements you know,patient outcomes for COVID-19 Here, I would say, quite a lot of good results and I stronglybelieve that here AI has a huge potential, right? In the battle against pandemic in general andspecifically COVID, right? However, there are several limitation factors that slowdown successful practical deployment of this AI based tools You know what is interesting? One ofkey challenges which AI experts are reporting is a lack of access to sufficiently large data sets fortraining and external validation of our AI models.

We generated a lot of data, but still, thereis a challenge for AI experts Without appropriate data governance and protocols for pandemic,one can hardly address this problem There is also a need to educate AI experts on aregulatory landscape, providing the deployment of tools in healthcare and actuallyclinicians and other experts should work with data scientists in multidisciplinarycontext to address concerns over data collection and privacy I believe thatthis cross-functional teams, whether expertise in medical data collection should support AIscientists who design this training algorithm and hand over to healthcare institutions totrain models locally.

And addressing these challenges will ultimately acceleratethe transition of higher research into practical and useful solutions forcombating pandemics, exactly what Sudhir reflected Thank you, Alexey Next question is related toour expectations of AI, and we know all the cliche that robots and AI will substitute usin one year or two or three whatsoever, which means that I think we are definitely wrong withour basic expectations when we speak about AI and on other hand, as we all know, in many regards and casesstill the infrastructure for proper AI is still not ready yet, right? From internet speed,from computing power and so on.

So my next question would be: are really our expectations of AI wrongand not aligned with the current possibilities and capabilities? Sudhir, maybe we can start with you,because I feel you are very strong in that regard? Sorry, I was unmuting myself Thank you, Dario I thinkthis is an excellent question, with respect to you know, the state of AI as wesee it right and you know the state of you know AI.

today versus you know what is actuallyavailable in an industrialized context is where my focus lies So I think youcan do a lot in a laboratory setting, but when you go into a factory setting, where AIrecommendations are actually driving you know the business users, that’s where things you know are not as they should be, right? And the difference is, like you pointed out,that you know, there is a certain view that AI is going to magically solve allthe problems or AI is going to replace humans.

I think these are two topics right,which are to be addressed right, from a education and training perspective, right? SoI absolutely agree that you need to contextualize your expectations of AI based onmany things One note is the state of your enterprise, the state of your organizationand the maturity of your organization is a key determinant of what output can you derivefrom this technology, because it is, I think Philippe mentioned, it is a combinationof process, technology, people, all of that coming.

together So its not a one dimensional thing So that’s one determinant and typically, obviously, the bias is that I will start with the AI solutionand then it becomes the next new shiny thing and then you have the hype cycleand then you go with the excitement and then the slope of the solution went You typically follow the hype cycle as you go towards industrialization, right? So the secondkey topic comes from our previous conversation.

on infrastructure If you have planned forinfrastructure, not just on the technology side, but on the capability side So I giveyou a brand new car, but your users are unable to know the functionality of the brand newcar, its going to be a problem, right? With respect to how you use the brand new car So thesecond topic, which is also a determinant on your expectations of AI, is do your users know,apart from managing expectations on what the brand new car is, do they actually know what to dowith the car? And that’s a big topic.

Again, from a training perspective, thoseare two key areas that I am seeing in my daily life that, you know there are folks like Alexey,who are actually business advocates champions, who are coming forth, the folks likeOrchid who are talking about the possibilities, but where the connection happens is a goodmix of setting the baseline with the community that is using it So these are two topicsI would say, are very key to position your expectations of AI and the last one, with respectto people losing jobs is another topic that always comes up, and I would say we are5 years away or 10 years away from autonomous.

systems You will see thecars and autonomous cars, you will see autonomous rockets, you will see all of those kindof evolving and then autonomous applications in the life sciences phase probably 5to 10 years, where you’re going through the basis of the industry So my note isits going to be augmented intelligence and not artificial intelligence for next decade Thatsmy response – Thank you, Sudhir, very meaningful.

Alexey? There are actually different areas of application,with quite different adoption level and satisfaction of end users If you look at RNG AItechniques, even now, I applied across whole drug development process to speed up drug discovery,to build predictive models as well as I dont know, discovery of an identificationof biomarkers, right? Here, I would say that we have very solid progress right and ourmethods aligned with our expectations.

Here we have good and solid results Then whenwe are talking about commercial, its important first of all to demystify AI and when you dohear the word AI, its pretty broad term, right? And what is it? Any systems mimicking, I dont know,human decision making processes are AI, right? Its ability of a system to interpret data, learnfrom these data sets and maybe achieve some specific goals using obtained knowledge, right?If we expect that we put all our data in one.

algorithm, one engine, run it and as a result we know exactly where to go with our business, most probably existing technologies are veryfar from it There are several enablers of digital transformation and pharma, I dont know,like AI AI is one of them And in a nutshell, AI and machine learning techniques enable betterusage of the data which we are collecting and efficient usage of the data brings positiveimpact on our services and customer satisfaction However, it can hardly substitute a human makingfinal decision.

We need to identify where AI is superior to other technologies and then the humansright, and what problems we solve and what success looks like Then probably level of satisfactionwill be higher Together with quite significant progress of AI applications andpredictive modelling, such as forecasting, it would be great to see more examples of descriptive andpredictive analytics cases, especially at the level of medical representatives They really change insplit between transactional and non-transactional activities while they engaging with thecustomers, right? That implies increasing level.

of complexity while we are moving from face-to-face, pure face-to-face like we had in the past to hybrid engagement Here dynamic analytics andreal-time recommendations are bringing a lot of value and I dont think that available commercialproducts are fully aligned with our expectations for these specific, for such use cases – Thankyou very much, Alexey Maybe just to add, especially what Sudhir mentioned previously, I think that on somejobs and vacancies we see a quite big pressure in overall, because we see, as an example in UK,we miss 5000 truck drivers, right? And on.

the other hand, we see Nikola, we see Tesla, they areproducing autonomous driving trucks and so on So I think that some things, this urgency will justaccelerate and investments will even increase in that segment and regard And even, of course,the regulations will also become much, much better to favor such new way of technologies, right? Orchid,same question for you Are our expectations wrong with AI and I believe you really have some goodinsight, because you’re working with many, many.

different pharma companies? – So its, I wish I had gone firstbecause Alexey and Sudhir stole a couple of points I wanted to make, but I really want to pick up onthat So I am not an AI expert, Im not an analytics expert I spent almost 30 years of my life beinga user of these, well actually 30 years ago there were no tools similar to this, but lets say thelast 5 or 6 years we had CRM, but Ive spent.

a big chunk of my life trying to understand and,actually 8 years of my life trying to have sales folks and sales managers embrace toolslike this So from my perspective, I really, what really touches me is what Sudhir saidabout the two things that are kind of off on the expectation of AI One is that its going to solveall your problems and the other is its going to replace people So I didnt leave pharma to goto big bad tech to replace people, because already pharma has a funny reputation and I did not wantto go to the bad side again you know.

Ill ingest But the reality is, what gets me excited andwhat is exactly what got me excited when I was on the pharma side, which is how do we humanizethese things, whether you’re actually selling a drug I spent a big chunk of mylife selling in the MS space and having my team talk about the person with MS, not apatient with a mess, but a woman who cannot go up.

the stairs with her kids, because shes 25 yearsold and she cant function because of her MS And when we changed our selling model to that versusfeatures and benefits, our sales went up and what I want to do with AI in our company is exactlythis, because I think that AI is an overused word I dont think I understand it fully, I have alot of friends who I go to, who are super smart.

in this area, who educate me all the time andremind me that what were doing is not true AI, and I actually read a funny quote the other dayaccording to Googles CEO, Sundar Pichai, AIs impact will be seen greater than even of fireor electricity on our development as a species I think that’s a huge stretch for me I donot agree with that I actually think AI is you know its really the interface rightnow between humans and machines and if we can augment whats happening right now, that’swhere we need to go.

And I think, to Alexeys point, AI in the other side of pharma, which is R&D, whichis clinical therapeutics, has taken off and doesnt seem to have as much resistance to it, I think AndI think its because when you’re in a lab trying to improve your genetic pathway to a target, you’dont have humans standing up and going that’s going to take over my job.

I actually think herewere talking about, truly talking about possibly replacing some jobs, so some of that is true, Ithink Sudhir, to say that were not replacing humans I think that to some extent wewill replace some of the things that humans do and hopefully stuff that we dont like to do Soin our world for instance, the AI analytics that the predictive power of AI analytics isreplacing those years I remember toiling over excel spreadsheets, which unfortunately hasntchanged that much Pharma is still using those excel spreadsheets.

In fact, I think Im using thoseexcel spreadsheets So those are the things that we need to kind of upgrade on and I think, in termsof problems, someone said earlier that you know I think it was you Philippe, in terms of the peopleside is over here and sometimes the business side is over here I obviously see that and now, inmy world, as I deal with many different pharma companies, some of them I worked at before and Ithink the issue is sort of trying to bridge that gap, so when you say to build acapability, absolutely, but we should be building.

things with AI that dont require a ton oftraining Some of these tools that we use are very clunky and theyre ugly and the reps dontlike using them and they wont use them, and they will put garbage in, garbage out So for me, itslike lets build some beautiful tools in pharma Why cant a tool look like an iPhone andbe gamified? And so that’s where I have a bit of.

a bias, because I dont think that pharma isin that mindset to create beautiful tools and those are things that, when I say beautiful, Isay it kind of light, but I do mean things that a rep will pick up and use And maybe we dont haveto build everything all at once, maybe you can pick up some of those things and you know our companieslike ODAIA are attracting a different kind of talent than a Novartis would, for instance, or aSandoz would And I think its just because these people are built to create these things and wecan gamify a lot of this to the patients benefits.

So this is where I think, I agree there’s along process, the data has to be clean, there has to be systems, but if we wait too long till that’shappened, I think some of the smaller companies possibly will buy off the shelf and jump ahead,because its not The fine-tuning maybe can happen over time and it doesnt have to happen in thefirst step – Absolutely Thats natural evolution and selection, I would say.

Philippe, same question?- Yeah, its quite interesting coming after all this and coming on the back end of four already pretty interesting points of views, but actually, I can draw some parallels and make a few connections here So the first one is, Alexey,you talked about demystifying AI and certainly and then you alsomentioned you used the term use case.

When we came into the pandemic we were already doingAI type of applications and Orchid you were referring to the Salesforce, so that was the very firstapplication we saw in commercial machine learning type approaches, with what we called internallysmart suggestions So you know, we would have have the rep accompanied by thesmart suggestion engine that would say hey, heres what you should do next when youmeet with this customer based on the history we have in terms of interactions and content thatyou showed him And that was really that I would call it if you’re going to gothrough a crawl – walk – run – fly type of approach to.

establishing AR, that would be kind of yourfirst use case and the one that everybody established And then on the human eyeside, which you referred to Orchid, which I think is really important too, we said to the reps at the time, think of it as a golf caddy Youre going to go out, you’re goingto play a round of golf and the golf caddy is going to make some suggestions on the club, but inthe end, you have the final say.

So if its if its a seven ironthat hes suggesting, but that’s not one you want to do and you want to go for a six well, that’syour choice, you know you can go out and do that And that worked fairly well and that the time toolwas interesting is the AI would learn both from what suggestions the rep took and the ones thatthey ignored and you continue to build on that So that’s the very first use case and thenthe pandemic comes along and all of a sudden.

were saying well, wow, we need to be able to applythis to digital and all our digital engagement We need to go into really omnichannel and youknow we need to have AI applied to omnichannel, because our marketers, our people are setting up our omnichannel campaigns, its getting complicated for them So theyre the ones who really needhelp, but nobody had any really real solutions.

And maybe they do exist nowand certainly were starting to look into that So that was one of the very first use cases,we started asking our data scientists to start developing saying hey, start thinkingabout it more from an omnichannel digital first type of approach, but neverforgetting that omnichannel does include the sales force Well I think this is really onething that you know, we need to be able to move kind of that, from that scale of single channelor rep-driven type of channel to something as.

much more omnichannel in nature So that is you know one way of looking at it I think the other thing I wouldthink of is what about the content So when we were thinking of channel engagement,we were thinking of channels and we werent really always thinking about content and contentconsumption And well talk a little bit more about that later, Im sure its one of the questions weget, but its also understanding content, because.

the content they consume, that’s where you get alot of insights on what customers really value And so I think that is also the next frontier,is really doing a proper job of understanding the content and what customers value, so that wecan continue to return value to the customers Thank you, Philippe Learned a lot alreadyand next question: we all know that culture and change management and changing people is, ofcourse, one of the greatest challenges, right? But.

my question would be: why people usually dontwant to do what AI says or why they dont want to listen to AI? Philippe, lets continue withyou, if you agree – Yeah, so I definitely, you know mentioned the golf caddy example, whichwas one way of saying to people look, you know look at it from a you know a different perspective,this is not here to tell you what to do, it is here to make a suggestion.

So you really needed, cant put enough emphasis on the the human approach I think the otherthing is, which I also mentioned initially, is we needed to spend some time and I think Sudhir,you went down, you were going down the education path too So we needed to educate, so weneed to educate our data scientists, I mentioned this, you know learn that data scientists need tolearn on how to work with pharma marketing and I specifically say pharma marketing, somebody elsealso talked about regulations and that we need to be able to work with within our commercial space,and then the other way around.

How do we get our marketers to be much more data-driven, to think inmuch more of an iterative fashion, shorter cycles, not be thinking I got a whole yearto optimize my campaigns No, no, no, this is internet speed We need to be able to do it, if not youknow overnight, at least within one or two weeks So there’s a, you know, and again, thinking peopleprocess So we adapt our people, we educate them, we adapt our processes I think agileness contextis very important and certainly, Im sure were not the only pharma company that’s invested inagile pretty significantly in the commercial space,.

so there you adapt your processes so that peoplebeef are thinking more about that iterative type of approach, and then, you know the finalpiece, of course, is the technology, which is about being able to, and I think there wasa really interesting point to made by Orchid, about the UX, you know the userexperience So its nice to think about the CX, for the external customer, but what about the userexperience for the internal customer? So that yes, you spend time on educating them, but that time toeducate them goes down significantly, because the.

UX is better, its easier to use, its somethingthat you can exploit and people can be efficient without tons of education So that is also aninternal focus now, is not just thinking about the external world, but about making our internal worldalso better from a customer experience perspective Thank you, Philippe Alexey, lets continue with you Youre on mute, Alexey – Sorry, I think thatthe level of trust is not exactly the same within different application scenarios and, as I mentioned, adoption of AI, for example, inRNG is is better than, for example, in commercial.

when we are talking about for example nextbest action for sales representatives There are a lot of well-establishedmethods in medical image recognition, right? And all these technologies are working prettywell and well accepted by healthcare practitioners The starting point for better AI adoptionis around bringing a higher scale I think that, and its joined as a function of data, systemsand the culture All data need to be consolidated around one platform and theyre well segmentedto be used efficiently Its critical to have, I would call it homogeneous,you know integrated information and technology,.

something like, I dont know, essential record forcustomer data, right? Or transparency around impact of communication channels and use Unfortunately,we have a lot of legacy systems you know and we do not have any opportunities to build everything from scratch, right? Therefore, this topic should be addressed somehow Equally important and we mentioned this is a cultural perspective Theright culture starts with a common agreement in the company around data-driven decision-makingprocess while approaching business questions and this is quite important, because if you relyon data then we have, we should have a proper.

attitude you know, we should, instead ignoring data, we should try to understand interpret this data And its alsovery important to ensure collaboration between AI experts and business functions Philippementioned around the marketing and data scientists, for example, I think that our colleagues in RNGfound somehow a common ground, they are using common language, they are able to talk todata scientists Exactly the same should be done should be done by us in commercial.

And bythe way, as with all other technologies, we need proper support for AI from the side of technologyand organization So even 1, 2 years ago, we probably didnt have data scientists in our teams By now, all these people are kind of part of part of our teams, right? I thinktechnology, data and culture in terms of organization and proper attitude to data – So theseare the foundations? Thank you, Alexey Orchid, lets continue with you – Yeah, I love what you saidAlexey.

It brings me back, I feel like Im an like my world is so much older Theredefinitely was no data science when I was in pharma to that extent and that’snot that long ago and I still have people come to me and say what does that really mean? Isthat an engineer? Is that a software person? Like people within organizations that we talkto still are very confused about that role and I think part of the culture change isexactly what you’re saying, is to educate people on.

what it means to be making decisions basedon data and intuition, not just intuition and again, going back to your example, which Ihonestly hadnt thought about until now, which is why is it that AI is moving faster on theother side of pharma Its because there’s no person putting up their hand going I actually thinkyou know the DNA goes this way versus this way,.

but we have humans that have been doing this fora long time and sales is a very, obviously, a very touchy-feely, intuitive, but also analyticalprocess and I think what I would love to see is that, just like we brought in the medicalside to teach us about, so medical affairs plays that role in pharma Interestingly, I dontactually have a background in science either and I survived in pharma for a very long time, butI used a lot of those people to help me understand.

what is a T-cell and a B-cell and all this stuffin my language, that would help me talk to an oncologist I didnt need to understand everythingand so, but there was definitely that translation that happened between medical and marketing,medical and sales to allow us to understand the need for that Being evidence-based when you’respeaking to a specialist Its the same thing where I think, as you mentioned, there’s,everyones coming in, there’s a lot of digital.

folks, a lot of people dont have the experienceand theyre becoming you know digital people, and I think, we as a company actually at ODAIA, werereally trying this hard Were small, but we are actually partnering with many different companiesto try to educate on this front, as well I think obviously, internally there’s a lot ofeducation to be done, because you have the experts internally, but there has to be some kindof understanding by a sales rep even what is AI and its not going to take away your job.

Beforeall of this comes down and says hey, use it and I think that’s what the gap is right now, thatculture is just happening really fast I think people just need to embrace it Im frustrated whenI speak to marketers I used to do marketing for many years as well, where theyre saying you knowwe cant talk to you right now about analytics stuff or AI or digital, because you know werein the brand planning process and Im like well isnt this part of your brand planning process andwhy are you using digital only at the end of your.

brand planning process, when you’re spending moneyon the digital stuff? Dont you want to know what the value, to Philippes point, like the omnichannelsall have value and dont you want to look at you know what these things are costing you before you’decide you’re going to do this, this and this and this? So I dont think an average marketing personreally, right now is really seeing the opportunity of learning how AI, whatever you want to call it,AI, ML, predictive analytics, how this can be a real huge resource in a competitive way, ifthey can launch a product knowing exactly you.

know which customer they should go after, with whatproduct, with what message, with what content, with what channel Those are all things that marketershave been saying for years and years and years Now you have a tool to be able to do that, but Idont see marketers ask for this, even if its not available, they should be asking for it And I thinkits up to us, as leading this conversation about this and demystifying AI, to bring thoseeducational programs in, so theyre very accessible and get people to start using someof these concepts and words and maybe.

its kind of like a new idea I have now, maybeI need to start a new business, but its talking to the average user of AI The other part ofthe culture, I think has to do with, I dont think companies, everyone talks about agile I thinkit means, like AI, it means different things to different people I think agile in a tech companyis very different than agile in a pharma company I would say shrink it by a thousand in termsof time, the kind of stuff that we do overnight, Im.

pretty sure pharma cant do overnight So whatIm saying, the reason Im mentioning this is, if you want to be agile, Ithink its not a bad idea to think a little bit like these little companiesthink and some of that is possible within small pilots, for instance We haveproducts where we need to iterate with a customer, we need to go back a lot and design it, or ifits already designed, we still need the users.

to come back to us and validate it Pharma hasa tough time with that, because again, the culture is very much we go to a consultant, we give themall our stuff, we give them our data lakes, we get them to get in there for two months and they comeback, they do a presentation and then we take that project and we do something with it and then wego back to the consultant if we need their help.

I think that its possible to use some of thesecompanies like ours to accelerate that thinking, bounce back and forth and get your peopleto start iterating more, so than you know putting things in business plans for a year andthen re-looking at them the next year, because technically, if you use AI properly, you’really dont need a business planning cycle to be the way it is You can kind oflook at it in real time and make decisions in real time and maybe get away from that, youknow those three months where you guys are all.

hiding in these presentations So thatI think would be a really cool impact of AI Yeah, good point, especially the point thatwe have to educate about the purpose of AI, because I think that in many cases people dontsee the clear purpose Not just about AI, about many other revolutionary technology and processeswhich are just around the corner Sudhir, lets continue with the same question with you – Yeah, thanks, Dario I think I have two parts you know to cover on this topic right One is, you know others covered already some environmental factors around the cultureand also improving trust with AI.

I want to take up some basics, which is notthere, some of the basics with the people who make decisions based on AI, whoare not data scientists There are two topics that improve the trust with the models One is abasic, which is correlation does not imply causation, right? Thats a really fundamentalthing that people miss and that’s very important.

And the second one is whatis your model performance Whats your AI model performance, whats thedrift there? And what is the sensitivity and specificity of a model? Theseare things that help you improve the explainability of your model and build trustwith the model for the decision maker Some basics, but it gives you a sense of understandingwhats under the hood, how does it work and what are the hyper parameters thatI need to get the best out of it There was a mention from Alexey on training data sets.

Its a loop, its a feedback loop The more you pump into the system, the betteryour model performance is In my previous job I was dealing with real-timerecommendations to cataract surgeons performing surgery If my recommendation went wrongfrom the software, there would be an explanation, the patient has to come back again and we have tofix it Its pretty serious So the other aspect that comes is the framework that you putand I want to expand on the framework piece.

to culture So the culture is very importantfor improving your trust and its a really good question therefore to connect the two, so onthe culture piece, I think there are two aspects which are very important The one aspect isyou need to be quantitative If you are having a culture that is more around people, rock stars andall of that, to move to a more normalized way of demystifying decisions and beingmore quantitative, this is what people talk about data driven decision making.

And so theculture has to feed into it Show me the data, show me what the model performance is,how you’re correlating your decisions and how you’re making decisions So this isone aspect that has to shift in your culture, because typically, when we dealtwith, in my previous setting, the message we were giving was you have a rock star performer, youhave an average performer, how do you bring your average performer to rockstar performance?And so, you know you have to facilitate the conversation in your culture toshowcase quantitatively this is where you stand.

If your sales rep is using this, there isobviously somebody whos doing well in the team, there’s obviously somebody whos not doing well,but if you anchor the two on saying both of your performance is going to improve, your incentivesare going to be better, whatever they are The culture amplifies the use and therefore, thetrust, because you have to make friends with it And the second point I want to bring onthe culture aspect is, we had several speakers.

come in lately, talk about this and we say: it hasto be a cross-functional team It cant be a team where you put a tech person and you put somebodyand then the whole thing has really no grounding So I would say the culturehas to move from a pure tech or a pure pit to a hybrid one where there is a healthy mix ofcross-functional folks So in our case, we have a good mix of the likes of Alexey, thelikes of my data scientists, the likes.

of people who understand data and you know thedownstream users So it needs to be all of that to improve your culture, to build trustwith that So very nice question and I really enjoyed this question, a lot more as we dig deeper – So is there any, to break down the question any further,is there any simple advice which you can give how to implement properly AI, so that it makessense and purpose within the organization? Maybe Orchid can also answer first on that, because we allknow that you work with many pharma companies, so.

you have this different angle, this helicopterview, right? – Okay So I mean, from my perspective, I think going back to what pharma always triesto do, which is put the customer in the center, and if you assume, in this case, the customer,lets talk about HCP as a customer instead of the patient for this discussion I think thatcentral view needs to happen in the way the data flow comes in, but also how unfortunately,the silos need to be removed.

By that, I mean if you’re looking at truly being customer-centricand putting your physician in the middle, that physician is a human being and doesnt justtalk to reps or doesnt just go into your portal, doesnt you know, hes doing things on Twitter,hes a patient himself, there’s all kinds of social things happening that need to be partof that On top of that, the medical team is having interactions with that customer from a verydifferent angle and sometimes directly through marketing automation and I feel that that’s thepart that’s really hard for pharma, and I get why.

its hard, because these are massive organizationswith hundreds of people in each of those silos But there has to be some kind of thinkingof when you pull in the technology, whatever it is, AI or not, puts that patient or physician atleast in the middle And that’s something that I really feel very passionate about that Imtrying to bring our company, to say yes, today were a tool for commercial, but wehave organizations that are coming to us saying we want our medical team in this platform, becausethey have more information and more data and were.

not utilizing this for the algorithm So how dowe do that? How do we do that in a compliant way, where of course, there’s firewalls between medicaland sales, medical and marketing, but there are there are ways of pulling that in So I would thinkthat one AHA! would be how do you actually, as you said, I think it was Alexey, to say youknow pull the silos apart, not just on between.

data science and the business, but put it inthe top of the physician and see what else is touching that physician, and sometimes, wecould use AI to predict behaviors of these people outside of what theyre doing with pharma andfor instance, what theyre doing on LinkedIn and how many followers, how many trials, publications Theres a lot of good companies pulling that data in on your competitors and that is somethingthat we like to pull into the algorithms as well, to bring you data that you dont have Thatsone thing, I think is an advice, because itll itll push you above what the traditionalviewpoint or the pain of glass that you’re.

looking at perhaps The other is to actually scorethese So we, you know, to be able to have a way to always prioritize in real time and havetriggers and scoring and that sort of thing, I think is really important, because then,the users, the representatives for instance, will actually see how that impacts their life and,to someones point earlier, becomes a process, a way.

that they actually behave So if they see the datashow that the priority of this position went from, I dont know, eight to three, and I didnt knowabout it and they act on it and that patient is benefiting from it and gets treatment faster,I think that’s very impactful and its very engaging for that fits the patient and for thatto rep Youll have your first champion right there So to actually have these metrics builtaround how these physicians or how your customers priorities are changing and how the actual usersare using that, I think that those lessons will be.

theyll just catapult into trust, I think, ratherthan just building it sort of in the system I think I mentioned this before, the other thingI think would be really good for us, that we take a lot of care in is design something that’s very,its kind of a jargon word – frictionless tools that people want to use Data that its not justthe data if the insight is fair and it takes the user you know half an hour to figure out how toput their password in, there’s a problem there.

Make it frictionless and make it easy Sothe user interface is very important, I believe And then have the feedback go in, so this cultureof, if you’re any kind of department going into that customer, even if you’re lets say in marketaccess, have that culture of them using that CRM the same way When I was in pharma, we had to forceother areas to actually use the CRM.

I dont think our MSLs were actually using CRM back then Theyare now, but you know, like it was just culture like I dont use that, that’s commercial No, its your book, its your bible So that sort of training as well, because all of this will build trust and will build the momentum to have a good system whether you’re building itinside or using capabilities from the outside Thank you, Orchid Philippe, simpleadvice to build up a proper AI culture? Ill try to keep it simple No, I think actually it isnt that complicated, at least to start with thinkingof this crawl – work – run – fly type of paradigm.

So data drives customer engagement or CX, gooddata drives good CX, and so you know I would first focus on that How do we integrate all thedifferent data sources that we have? And I think all of us know, whether you’re in technology ornot, that that is a challenge So all these digital data sources, you know the omni channelperspective on all the interactions and touch.

points we can have customers, so integrating thedata First thing to put in place The second thing and then thinking again to people – process -technology and how that needs to go lockstep So it means you need to progressall three of those dimensions at the same speed, otherwise if one falls behind, you’re notgoing to really be able to reap the benefits.

So one aspect would be okay, how do I get people to bedata driven? And I need to think about harmonizing my KPIs and measures and just do plain analytics,right? How can I even just start doing really great descriptive analytics andprogressing that towards predictive? And so a lot of groundwork needs to be done in terms ofharmonizing your KPIs and measures and getting agreement on those harmonized KPIs and measures And I make a distinguish, you know I distinguish between KPIs, which are really the critical few,and just a general set of measures, which can.

compose your KPIs and there’s quite a lot, I mean,I think we went to build up a library from close to 100 different types of KPIs and measures thatwe could use in the commercial space at one point Then the third thing, and Dario, Ithink youve heard me mention this or I certainly asked a question around it is, we realized that weneeded to do a better job around, especially when it comes to digital, tagging taxonomy So how do weset up a common taxonomy, globally, for how we tag our web pages, our url, our content and that way wecan really understand what customers value, what.

theyre consuming, help us be much more efficienton our marketing, quickly throw away what doesnt work andthen continue to focus on what does and that type of thing So if there’s going to be you knowthose three basic things that I would do to get us on the path to AI, Im not saying that’s going tosolve everything, but that’s, I think, a prerequisite or a necessary first step Its really on thosethings, integrating your data, thinking about that Harmonizing your KPIs and measures, that’s alreadya good step in terms of being able to be data.

driven, and then number three is that commontagging taxonomy and that’s going to be absolutely critical if you want to get any kind of insightinto what your customers want And then just be able to do some global roll-ups and understandwhats happening at a brand or franchise level Thank you, Philippe Sudhir, lets move toyou – Yeah, so I think I covered Orchid and Philippe mentioned, so key topics frommy side that we are focusing right now is.

focus on the capabilities instead ofthe technology and dimension So the capabilities of the enterprise, withrespect to whoever is making key business decisions How do we bringthem into the equation and clearly keep the value into focus Right now, you seem to think of AI as this other thing and if you bring AI and business value together, youare amplifying the impact of AI So I think when Philippe and Alexeymentioned about use cases The use cases have to translate into business impact and value Andthat’s the focus.

And so in a way you amplify your usage of AI and yourculture of AI, but keeping it actually in the back burner and improve your storytellingnarrative on what does it mean for the business And that’s really, keeping that in the center ofthe equation, actually scares less people away from whatever needs to be done to adopt andmove people to their core of how it is helping Thank you very much.

Alexey? I would say that we are talking a lot aboutcustomer centricity, therefore I would say that culture, it should be function of actually ourcustomers, right? And Dario, I would suggest to rephrase the question a bit I would suggestto think about something like what does successful engagement look like? You know, fromthe perspective of our customer and ourselves, you know sales representatives And I believe thatfirst of all, we are talking about natural two way ongoing conversation between HCPs and our reps,right? Therefore, high responsiveness to customer.

requests is critical success factor here Its partof the culture Taking into the consideration challenges with access to HCPs, rep should playa role of the kind of solution provider, right? They have a good chances in this caseto keep relationship and ensure that they still have personal touch points, ifits necessary on any types of engagement, to ensure that there is a continuous communicationbetween two parts.

Therefore, successful engagement is driven by ability of rep distillcomplex information into, so to say, simple messages and the ability to use efficiently,use this information to generate recommendations I think Orchid mentioned this, in real timeusing internal analytical engines Therefore, if we approach the question of culture from theperspective of final outcomes, ensure customer satisfaction, right? Then, its really importantto ensure that our customer, patient, colleagues.

responsive, they would like to bring solutionin real time and there is a proper support from back office around, from the perspectiveof analytics, from the perspective, I dont know, of medical colleagues, marketingcolleagues and so on right? I dont believe that it is possible tojump from where we are now, you know to fully customer-centric pharmaceutical companies,its very, very difficult, right? However, if we know our ultimate goal, final objective, you know to becustomer focused, customer-centric, then probably.

we would be able to find out our way to getthis objective And data systems and culture, these are to enable us toto achieve this final objective for us Thank you, Alexey We are left with two minutes,but we can extend it for another two, if the audience agrees and Alexey, you reallyset a stage for my next question, which is: what are the relevant insights which can leadto better customer experience towards HCPs?.

So then lets stay with you, Alexey – Look, once again,this is continuous discussion, natural continuous discussion, what we can observe over the last12-18 months? From a transactional sales process with transactional activities in back office,SAP, you know, with transactional planning, what Orchid mentioned, brand planning, I would liketo be the different one We move to something which is continuous, therefore we havecompletely different expectations around.

all functions, engagement, collaboration betweenthese functions and support for customer facing colleagues who are actually bringingall results from the whole company, from healthcare providersto healthcare practitioners, right? So that means if we are able to ensurethat there is a discussion in real time, recommendations to sales representatives or nextbest actions we are able to provide in real time, it is not discrete, then I think our customers willbe happy This is the first point And the second point, which is equally important, we shouldbe always in the listening mode, because I can.

see that pushed activities from, us a workingefficiency of these push interactions are less and less effective, therefore we shouldthink how to change our overall engagement process from push to pull, where healthcare practitionersare comfortable to come to us, theyre able to get right service and then it will be realpartnership It will be not one-way communication We are moving, I dont know, from productsto patients, we remained relevant for them These are a couple of elements I believe all andimportant for the success and appropriate.

customer experience on the side of physiciansor pharmacists – Thank you, Alexey Orchid, maybe we have one minute for speaker left since – I mean, somuch great stuff has been said on this already, and I go back to what Sudhir said earlier,you know dont try to solve all your problems with this Theres a lot of things todo and maybe the biggest impact you can make today is, exactly what we just said, which is let the HCPdecide where they want that information and how.

they want the information and try to make thatas seamless as possible And when you do it, try to be as personalized as possible And I thinkif you can think about that and have your AI predictions lead towards that, youll find thatyou make better decisions using data and the human approach is added to that, then you canwin through that.

So its really trying to think of it backwards, instead of pushing it down to the HCP I really like that in terms of you know, even from the patient, what do they want and design yoursystems around that – Thank you, Orchid Sudhir? Yeah, so for me, its the insightsrelating to business process performance, with respect to efficiency and effectiveness, arereally going to help.

So specifically like if you’re looking at channel performance,campaign performance, brand performance, right? Those are aspects that are going to helpyou have better insights about how you’re doing, with frequency and reach SoI would say, write all the elements of CX, the metrics that you care about you knowshould be driving you towards helping better reach and a better personalizedimpact, with respect to the end user So that’s really key here The metrics that you take into account from all the insights that you deriveare really going to key, to tap into.

the part of the customer universe that typicallydoesnt meet the segmentation criteria and also, are trying to improve affinityfor existing customers Again, there is a whole set of metricsthat you have to focus on, amongst all the insights that you gather and those metricsthat you’re focusing on should be focused on the process aspects and that will help and that’sprobably a great way to feel it to close from Thank you, Sudhir And final words, Philippe? – Yeah,Ill try to make it very quick So its kind of reinsisting on something that’s pretty much saidwhich is dont try to boil the ocean, really resist.

that temptation to go straight and say Imgoing to do omnichannel next best engagement, were even calling it next mix experience nowinternally, just to give you that omnichannel flavor So certainly, you know, dont do that Think big,start small, set up some experiments on some very simple use cases using that terminology I thinkyoull find that you know there’s a lot of very simple tools that can help you get started, marketmaturity, something that’s super important, probably.

working with Orchids company can get youstarted on some you know smaller use cases, but experiment and grow from there If that wasone piece of advice I would give, that would be it Thank you, Philippe And this leads me to the end Thank youso much for joining this interesting webinar! We will also publish it on our official YouTubechannel And once again, thank you, Orchid, Alexey, Philippe and Sudhir! Was an amazing conversationand looking forward to the next one! Take care!.