Krishna Cheriath: AI is Disrupting Clinical Research Already

Ram Yalamanchili (00:09.743)

Hi, Krishna. We're so excited to have you here today. I've been looking forward to speaking to you for some time and very happy that you were able to give us some time today. First, before we jump in, Krishna is a thought leader. He's been at the intersection of biotech, pharma, and AI for quite some time. He certainly has a lot of interesting thoughts on his LinkedIn if you haven't seen his posts in the past. So I highly recommend

checking it out and you know, maybe you can tell our audience a little bit about yourself and a quick intro on what you've been up to lately.

Krishna Cheriath (00:48.696)

Ram, first of all, huge thank you to you for inviting me and I've been following your journey and the Tilda journey along the way. Love the investment that you're making in the fluency space and trying to progress the industry as a whole. So kudos to you, kudos to your team on progressing the mission. In me, you have a huge fan. So with that as a...

As I said, let me just talk a little bit about myself. Krishna Cheriath I work as the head of digital and AI and clinical research in ThermoFischer Scientific. Prior to that, I was the chief data analytics and AI officer in Zoetis in animal health for about three and a half years. And prior to that, chief data officer at Bristol Myers Squibb. And then I like to say I'm a recovering management consultant from back in the day.

I'm also passionate about teaching. I work as an adjunct professor at Carnegie Mellon. My focus is executive education, and I do a lot of programs with, and very proud of the fact that several leaders have come out of that program and become chief data analytics officers themselves, and also sit on the board of three analytic and AI startups.

I think the best way I can capture my personal journey on this has been, I was born into a family of physicians back in India in my home state of Kerala. My dad was an army medic, became the director of health services for my home state. My mom...

is a gynecologist. She succeeded him as the director of health services. My elder brother is a practicing cardiologist. My sister in a dermatology practice. Uncles, cousins, all in the medical profession. I was the black sheep in the family and went and did electrical engineering and then, but somehow ended up in the intersection of healthcare and digital.

Krishna Cheriath (02:51.414)

And for the longest time in my journey, I've been in the business of supporting biopharma and biotech from a data, digital, and AI standpoint. And always that mission felt kind of a one degree remote for me personally. I mean, you do lot of work from an enterprise setting, but the product that you are bringing to the market is really medicines. And you're not at the tip of the spear, you're enabling the tip of the spear. But all of this changed.

as I was sharing with you with my own family's journey with serious illness. And when my wife went through a diagnosis of breast cancer during the COVID years, nothing prepares you for a diagnosis like that. And it was completely out of the blue. And then she went through a year and a half of a remarkable journey that one

I'm so proud of her. And I always reflect, like, if I wasn't that close, will I have the same courage and grace under that kind of a condition? And I would out it. She was amazing. And then that gave me a lot more insights into the journey that a patient goes through in a journey like this. And the net-net of this is that she's alive today, and she is having a great quality of life.

And she's with me and my family, mainly because of the medical innovations that came to the market over the last 15 years. And in many ways, so this journey became a lot more personal to me. And a lot of times, we can be our own worst critics when we work in supporting R &D and manufacturing, because we all know how much more progress we can make. But first, we have to start with how proud we should be.

collectively as an industry around the range of innovations that has come to the market. One of my favorite books is the Emperor of All Maladies about cancer. And the PBS series is a wonderful series as well. And the fact that we have made several cancers into a chronic condition and manageable disease is remarkable. so this whole journey around medical innovation, bringing these innovations to the market faster.

Krishna Cheriath (05:07.854)

and affecting patients' lives have become a lot more real for me and gave me lot more meaning and purpose to the work that I do in a way that was not there before because it has always seemed very abstract to me if I'm being very, very honest. And so now I feel very proud of the fact that I'm working in this industry. And whatever little contributions I can make to the progression of it, I count as a blessing every day.

Ram Yalamanchili (05:33.753)

No, absolutely. think, you know, I wouldn't call myself an insider or a planned entrepreneur in this space. Very much a deliberate but somewhat accidental. Right. And it's interesting that you shared a couple of perspectives here. For example, I have come to ask this question where I, whenever I see somebody who's like not that typical fit, right? Somebody who comes from a biology or a biomedical, that's sort of a background working in these industries and

really trying to push the frontier. I almost always like my default question is like, what motivated you? Is there a personal reason? And the personal reason, believe it or not, is something I've found to be the reason in many cases why people switch their carriers and actually like are here and motivated. they're like, the drive is not just, let me go do something, right? It's much further than that. And without getting into details, I think we both share a very similar

history on why we're here and why we're motivated to change it. And I found this to be a fascinating space where it sort of brings in that type of a talent to look at it and sort of try to do something different, right? So, given what you've said, I do think...

There is this like notion which I keep hearing, which is, like, you know, so much has happened in the clinical research space. So much technology has been adopted. And yet it feels like we've lost a decade. Because if you look at all the typical metrics, you know, I wouldn't need to tell you this, like, you know, whether it be how many trials we have enrolled on time, how many medicines we've gotten out in time, like, you you just kind of plot this out of the last 10 years. I mean, there's improvement, but potentially like not, you know, when you risk adjusted to the

cost or any other sort of metrics. It's just not there. You sort of very clearly see these issues, right? So I'm curious, first of all, do you agree with that? Do you see that this is one of the challenges which we are yet again facing? Because now we've got AI, now we've got all this new sort of technologies. And the question always comes back to, well, is it really going to make an impact? Are we actually in a path where there's something meaningfully different which is going to happen in the next 10 years?

Ram Yalamanchili (07:49.007)

Let me pause there. What are your thoughts on that? I'd be curious.

Krishna Cheriath (07:53.017)

Yeah, think the two ways I'll answer it is it's a classical, may sound political kind of a yes and no answer. I think if I look, if I disaggregate that whole question into two parts and say, the first part of it is the drug discovery aspect of it. And I've seen up close how much evolution technology

business process, reimagination, analytics, et cetera, has changed the arc of drug discovery. And how, from a target selection standpoint, the refinement standpoint, identification of biomarker standpoint that is data-driven, is remarkable, progression has been made. But you're right on the yes part of this is on the drug development side. I think if you look at

In many ways, I was talking to the of R &D and I said, it almost feels like we have developed these Lamborghinis and Ferraris in the drug discovery side and we are driving it through a single lane road with the potholes along the way. When you think about the overall drug development timelines and the process and how much cost it takes to bring medicine to the market.

On the first part of it, think we made meaningful progress. And on the second part, lot more to be done. And one of the things that has struck me since I've been in the space of that intersection of technology and this for the longest period of time is, number one, we may have reached the maximum unlock value of automation in this space.

Because historically, we have pursued automation and digitization in pockets along this chain. But I think to make meaningful progression, dramatic acceleration of outcomes, and whether it is timeline quality and others, I think we need to go beyond that. And that's where my excitement around the AI opportunity stems from, is because I'm convinced that we have made a kind of siloed progression.

Krishna Cheriath (10:01.058)

We have automated workflows, we have digitized workflows, but still we are not making decisions soon enough. We are still struggling with advancing some of the metrics and others. So I think that is so that you're right in your assertion around the saying that in many ways it feels like we haven't made enough progress in elements of it. But on the other side in the early stages of research, computational research and drug discovery, the progress has been amazing.

Ram Yalamanchili (10:30.991)

Absolutely. I think that we can't deny, right? We've had massive advancements. mean, even for example, Demis winning the Nobel Prize for essentially a protein folding model is just like case in point of what you're trying to say, right? Like the discovery phase has seen tremendous, tremendous improvements. Delivery, you know, obviously we've got more to do there. I think part of one of the things you're saying just from our audience perspective, right?

Krishna Cheriath (10:39.978)

Yeah.

Krishna Cheriath (10:43.736)

Yeah.

Ram Yalamanchili (11:00.675)

You've said we've made lot of progress on automation, but you're still looking forward to AI. I'm curious, what does that mean? Automation and AI, are they the same or are different? How do you think about it?

Krishna Cheriath (11:10.742)

I think if I keep, I get more at an abstract level to say, in order to make meaningful progression, if I just narrowed the conversation at the clinical trial space to say, what would meaningful progress lead? I am convinced that we need to go beyond automation of business process to much more adaptive and continuous ways of doing things. There is a lot that has to be resolved through

supplementing human effort with other mechanisms, reimagining business process with AI first. So I think that's where I feel we have hit the limits of automation in this in terms of how we have pursued automation in the silos. In silos, we have improved workflows, but collectively we have not made meaningful progress.

So when you think about AI, the first thing that comes to my mind is we should not repeat mistakes of the past by continuing to pursue episodic AI, if I can call that, in pockets. But we need to look at to say, if I start with a highly ambitious goal, and by the way, this was given to me by one of our sponsors who put this idea that, if our North Star is one day to start a trial?

one week to enroll a patient, one week to do database lock, and one day to submit the regulatory submission after the last patient last week. I may not see that in my lifetime, and there may be variations across therapeutic areas. But to achieve that, I'm convinced that we need to think differently about the potential of AI across this. My fear is that we will resort to our comfort zone of saying, I'm going to optimize this process, this process, and

throw air at this and lose the re-imagination of this at the table. And because the complexity of what you and I deal with is heterogeneous, so many different stakeholders, beyond the capacity of one particular stakeholder at once, I can do all of the tech I want sitting on the sponsor side of it or as a product and services.

Krishna Cheriath (13:19.458)

But if I cannot meaningfully advance the ability of a patient to find a trial, be able to afford and access the trial, and solve for some of the human conditions necessary to stay on the trial, what good is a patient recruitment deck? So I think we have to, as an industry, think beyond to say, yes, workflow augmentation and AI value unlock. But what is the larger mission that we are after, and how do we fundamentally disrupt that?

Ram Yalamanchili (13:48.559)

I see. know, kind of summarizing how you put it, right? I also have always felt, you know, the difference between automation and AI is two things, right? One is reasoning and the other is ability to contextualize. These two are uniquely human, human intelligence, right? Like in the past. And I think we've really shown now that these AIs in certain limited domains can be quite good at reasoning, planning.

Krishna Cheriath (14:02.829)

Yep.

Yep.

Ram Yalamanchili (14:17.987)

and then sharing the context. So I think now the question of like, how do you actually bring them to work well within a environment which is filled with humans? I mean, that's the next challenge, right? We're sort of exploring entirely new territory right now with what we've got. So one question I think is important to kind of point out or at least for me to ask is, used to it in a vantage point very few of us have.

Krishna Cheriath (14:28.109)

Yeah.

Ram Yalamanchili (14:46.157)

an organization, arguably one of the larger organizations in the pharmaceutical space, Google research space. I'm sure a lot of conversations revolve around ROI and ROI can be defined in multiple ways. But how do you look at ROI? how would you define, you know, definition as well as measurement, right? Like, and what drives you, guess, or what's driving your decisions on a daily basis?

Krishna Cheriath (15:08.686)

Yeah, and I think it's a very, very interesting question because we are in the tech space that we occupy, and especially in the enterprises. There is a lot of pressure around quantifiable, CFO certifiable value returns, whether it is top line or bottom line. And then there is always the question around what is that threshold?

that you need to cross in terms of the total return for it to be meaningful and effective, aiming the many investment opportunities that a company has. And that is an age-old problem for tech investments, and it is true in the AI investment as well. But I think what I have been reflecting a lot in this space is to say, if I am supporting the implementation of a capability that makes the trial execution faster by a day.

may not look very meaningful from a larger scheme of things from a company perspective or an organizational perspective. But reflecting back on that personal journey I went through, imagine that innovation being available to a patient one day sooner. That is an ROI that for that patient is worth so much. So the one thing that we have to be really conscious about in our industry is that it is easy to lose sight of

what the larger mission when we get into our day-to-day battle. And so when the definition of an ROI, have to, of course, each of us are working for for-profit enterprises. have a fiduciary responsibility to our shareholders, our customers, to our colleagues. All that is true. But at the same time, we should not lose sight of the fact that at the end of the day, what we are here about is to make that next big

medical innovation faster to the ultimate patients that deserve, that need it and is waiting for it. That can be a throwaway line in many situations, right? As well as a check the box line that we say. Qualitative saying, and it is very hard to define. So I think it is one of those cases where it is not an either or equation. I think we have to look at the and equation of yes, in order to have these investments, we need to have good ROI.

Ram Yalamanchili (17:12.879)

It's like a qualitative versus a quantitative type of a thing, right? Yeah.

Krishna Cheriath (17:29.74)

defined ROI in terms of time, cost, revenue generation, productivity optimization, all of that. But at the same time, we also need to measure that human ROI and then put that also as one of the decision-making criteria, which brings to a secondary point around this, which I've been thinking a lot about is, when you think about the space that I work in a lot, which is a clinical trial space,

In order to make total progression around the larger goals of accelerating trial timelines by 50 % or 70%, it is impossible to do that from the context of just one organization or a couple of organizations in the mix. There are so many nodes in this. Because whenever I was talking to a tech startup and the entrepreneur came from a completely tech background, and it's always fascinating because

She comes into this industry and saying, the hell are you guys doing? I this is what is this? And I'm sure that you must have gone through a similar journey.

Ram Yalamanchili (18:33.423)

Yeah, we've all asked this question, I'm sure, coming from elsewhere. Like, what are these guys up to?

Krishna Cheriath (18:38.54)

I think part of the reason is Scientific discovery is hard. There are barriers that have been put into the clinical trials for a reason to make sure that at the end of the day what we are coming out to the patients are safe and the efficacy is there. And then we are able to make that at the right balance for the human condition. So the barriers are there for a reason.

But what is undeniable is the complexity of the ecosystem that surrounds innovation is by far the most among many of the industries that I have had direct experience in. So if you're going to advance, for example, patient recruitment, I've been part of so many different conversations with many startups and others around EMR and real-world data-driven patient recruitment. Yeah, that solves 30 % of the problem. But that is all going into sites that are at varying levels of digital maturity.

and overburdened with understaffed and so much things to do. So it is not just a question of an identification of a patient, but it is a question of how do you convert that patient? How do you ensure that the site is able to act decisively on it and effectively on it? That alone is also not sufficient. The patient who is ultimately maybe going through so many different economic and other logistical conditions.

And even if you have done the best job from a tech perspective to identify that patient and you've done the maximum possible from site optimization, you still have a human at the other end. And that human may have issues like, I don't have somebody to take care of my kids, or I have an elderly person at home. And this trial, especially this is true in oncology trials, this trial is so far away. And if I don't have support, I cannot go there. I don't have the economic wherewithal to support this.

If that is not solved, then yes, all of this tech innovation from a patient recruitment standpoint is not going to yield the ultimate outcome, which is more throughput. So it's something I've been throwing. And then the last point I'll make on that is, then the question is, OK, how do you make meaningful progress? And what can you do? And I think this is where we need to have more evolved.

Krishna Cheriath (20:55.456)

investments and again if a socio entrepreneurship if I would say like just traditional tech entrepreneurship alone is not sufficient to make advancements in this there needs to be things where the profit margin may not be there but there is other socio economic value and the socio entrepreneurship needs to happen and I've been talking to a few people around this idea to say how do you promote that so all of this to say if you're a tech entrepreneur like you are Ram I would say

absolutely do not be disgusted or disappointed by the pace of progress. Recognize that you're in an industry which is highly complex, so many different stakeholders. Meaningful progress doesn't come in the same stages like in a B2C world, but each day matters. But we do have other things to solve for to make a radical shift.

Ram Yalamanchili (21:48.633)

Yeah. And, and I think you bring up an excellent point for entrepreneurs like myself and others trying to break into this industry or trying to make a difference here, right? I think the timelines are definitely something I've noticed to be longer than, than what you would normally expect in any other industry. You know, I've built a, a, a large company in the security space and primarily serving regulator industries like FinServe or, you know, traditional healthcare.

And I find this one to be much harder to crack, right? From timelines perspective. So I think I certainly respect that. And if anything, I've learned over the years that, you know, asking the why is like a really important question. why, like there's a reason why these are the way they are for decades, right? Maybe it's incentives, maybe it's safety, maybe it's something else, but the structure exists and the ecosystem exists for a reason. And maybe there are ways you can like sort of attack the problem in different ways, but.

But certainly respecting the problem and the ecosystem and the way it's set up, it's really important. I think that also leads you to become a better product person, right? Like you can actually make better product decisions than, know, GTM, like go to market. I think from an entrepreneurship perspective, I really felt like those are the type of things I tell other entrepreneurs working in our space, which is like, you you have to be patient, you have to understand the problem, you have to understand why it is the way it is. And it's not because they don't know. It's certainly, everyone knows all the...

usual stuff, but there's a reason why it's the way it is, right? And then, and then see what you can do from there on. So I have a sort of a question, which I think digs a little bit into your past. You know, you, you, you, you're a senior executive. You've been in the space for some time now, and I'm sure you've seen waves of different technologies come and sort of hit the, the market, right? The one I have, I guess, studied and,

really dug in would be the prior way of digitizing what would otherwise be all paper-driven processes. So we've had many enterprise systems which have come up specialized for research, like ERPs, which are specialized for our industry. So, know, the plethora of three-letter, four-letter acronyms out there, CTMS, EDC, TMF, ECOA, like all this other stuff, right? So my curiosity is that

Ram Yalamanchili (24:11.747)

We're agreeing that they've made some time, possibly not where potentially the ROIs may have. I'm sure at some point somebody did the ROI math and looked at both the quantifiable as well as the qualitative aspects of what we just talked about. But I am curious, what's your view? How do you think about the next 10 years or what's happening with the whole AI wave right now? Are we sort of maybe in a similar situation where we end up in...

We do all this ROI math and essentially nothing's really changed or like do you have a different view on this?

Krishna Cheriath (24:44.256)

I think we may be at the cusp of a fundamentally different pivot in how we look at enterprise tech. And I think if I go back on the evolution of whether it is starting from on-prem applications to software as service and others, there's a...

a constant theme of we took functions, business processes, and then automated it and put it into a standard package to say, to use the overused cliche from back in the day. You can have any car you want as long as it is black and it's a model V. That's a similar mindset we had in the evolution of the enterprise software applications is to normalize it down to a

standard flow and then put a shell, which is what we call an application over the data. I think what I'm most intrigued about is a potential disruption of that whole way of thinking with AI. Because if I fast forward and if I can be a prognosticator at the risk of being an astrodemons around this is I think the whole software application industry is going to get disrupted.

And I think if I look at the future where there is high quality, highly trusted data, but each of us as knowledge workers will have our own custom way of interacting with that data to make our decisions and actions through an intelligent AI layer. This could be AI agents and others. Why should each of us be constrained into a standard box of a business process workflow is a question that I've been asking myself.

Because each of us interact with technology differently, what we need at the beginning point of things. Why should I go through a standard? Just as a simple illustration, why do I need to go through a standard navigation before I get to the information? Why do I need to go through a depth by 1,000 dashboards to get to the information that I want? Why do I need to log in and then go through a standard set of screens to get to the information I want? I think that model of this, to me, feels like, you know,

Krishna Cheriath (26:58.094)

I am old enough to remember the phones which where you do that dials. saw a rotary phone. I saw the fancy or interesting YouTube video where two kids were given the rotary phone and they were given a phone number and said, hey, use this device, figure out and call this number in five minutes. And these kids were trying to figure out what the hell is this and what do I do to make a call?

Ram Yalamanchili (27:07.033)

The rotary phones, right?

Krishna Cheriath (27:27.95)

It looks so antique today. I feel like the software applications of today in many ways feels like the rotary form. I think a fundamental pivot I'm expecting to happen is that, and this is not to say all of that will be destroyed overnight, but a progression will be made to a reimagination of our computing space. And I call it the human AI workspace, where a human is able to interact with the data and be able to do the actions in a way that is meaningful to him or her.

that to that customized journey. And I think AI makes that possible. Today, I can ask ChatGPT to plan my vacation. And I did a family vacation last year. Every year, we pick a spot. All the extended family runs in Airbnb. 80 % of what we did was a ChatGPT design vacation. And that is a custom experience for me. And I think that idea that

We don't need a software application layer. We need data and we need to have an intelligent fabric of AI agents and I can have my own custom interaction so that I can make the fastest possible decision and the most effective action possible. That is doable. That is going to happen. And I think that will also will require all of us to invest heavily in our own personal fluency as well. And that is going to be the differentiating factor.

Ram Yalamanchili (28:46.819)

This is a fascinating area. I'm very passionate about just what we are talking about because it of revolves around my work as well. I've always thought about this analogy of learning is, or learning or even interfacing with somebody, if you think about just throw away all the tools and just naturally do it, We as humans, we have a couple of ways to do this. You can write everything down on a paper and then I read it and then I gain whatever I need to gain out of it.

or I can write something back to you and then you can respond back. So this is kind of like where our interfaces are, right? With software, where we're sort of in a very defined mode or button clicks or whatever in workflows for defining what we need. But I also see a much higher order learning where you can show me what you can do and I just sort of mimic you and I learn based on just by watching. some certain types of skills can only be done that way. Like there is no way I can just learn to ride a bike or

Krishna Cheriath (29:39.95)

Mm-hmm.

Ram Yalamanchili (29:43.171)

you know, learn to swim just by reading a book and then just jumping into the water, right? That's not going to work. it's essentially like, I think we are definitely in a paradigm where the type of tools and the type of technology which is available to us will greatly expand the type of work and type of learnings and type of added value we bring to the table. It'd be very hard for us to sort of pointify what exactly and which direction we're going to go. But this does bring me to the point which you brought up, which is on the fluency side.

And, you of course we're in this podcast because we both believe in AI fluency and we think this is a sort of like an imperative that we should talk about, given what we're seeing. So I think the question I have is, how do you think about or what would be the advice you give someone who's in our industry who's basically, you know, watching and they're like, you know, this thing is moving really fast. There's just a lot of talk around AI, lots of automation, lots of, you know, AI driven agents and whatnot.

Krishna Cheriath (30:35.374)

and

Ram Yalamanchili (30:42.413)

And I think there's a mix of fear and uncertainty. And I feel there's also a lot of a new cycle, which is not helping, right? They're taking it in the wrong path, is kind of my belief. And I think I can argue there's many reasons why it's the way it is. Again, going back to the why, right? Like, you know, there are incentives aligned on individuals and companies to sort of, you know, promote this kind of thinking.

But I'm curious, what's your thoughts for our colleagues in our industry who are essentially looking at this and thinking, what's the future going to look like?

Krishna Cheriath (31:20.854)

I think number one I would say is that this disruption is real. And compared to any other waves of disruption I've experienced over the course of my 30 plus years of career, I think this one has moved at a speed much faster than before and much more volatile than before.

So think there was a little bit more of a linearity of progression in the past disruptions. But this one is kind of hard to predict, but it's moving at a faster pace. But disruption is real. If I start with that notion, it cannot be ignored. It is not high. This is real. And that is reflected in, when I look at one of my favorite websites and newsletters that I subscribe to is called the Pessimist Archive.

Ram Yalamanchili (31:50.703)

So it cannot be ignored, right? So this is real, what you're saying.

Krishna Cheriath (32:06.648)

very curiously titled research. What they do is they go back and look at all the technology introductions of the past, and they kind of look at what was the zeitgeist at that time. One of my favorite examples is back in the day when airplanes were first introduced, the headlines on one of the major newspaper and the editorial said, we don't find any practical application for an object that flies.

And even the introduction of Sony Walkman, and there were all of these news items around how it is going to fry the human brain and then make zombies of us. All of this to say that we tend to be, as a society, tends to be over optimistic around the impact in the short term and underestimate the impact in the long term. And with the AI, I would say is that

there is definitely going to be disruptions for every kind of jobs that we do. Now, the question is, what is the degree of augmentation and by when? And that is hard to predict. So if I'm somebody working in this industry, the first thing I'll say is this is roughness is real. And I owe it to myself and to my family to take this disruption seriously and have to make sure that I the basic fluency around what is it? How is it relevant?

how it may be applicable in my industry and where this is headed. Number one. Second, I should embrace volatility. Because this is not going to move in a linear fashion. The things that you may learn today through any kind of programs, maybe 30 % may remain constant, 70 % may remain constant, hard to tell. So you need to embrace this and embrace volatility, which means that it is going to put a premium on you being at the

tip the sphere and constantly kind of looking at what is the incremental things that are changing that may be relevant for my industry and my role. So it is not a one and done in terms of just let's take a training. Yes, basic foundation, but we need to be constantly evolving. But then the organizations that we work for has a responsibility. So I view it as a 50-50 equation. 50 % is the individual responsibility. I need to make sure that I have the skills to.

Krishna Cheriath (34:22.708)

survive today and tomorrow in this industry, and then I need to keep myself at the edge. The other 50 % is where organizations need to reimagine the learning experience for their employees. Because if I look at every organization and if you imagine that every role is going to be AI augmented, there is no way you're going to recruit away the problem. You're not going to be able to recruit.

a brand new AI-fluent workforce for all of the roles. It's not going to happen, which means you have to upskill the workforce. And this is where I think people with titles like mine have made a disservice to the industry. Because historically, if I just look at data literacy and analytic literacy programs, and I've been part of several journeys throughout my career,

And as I was sharing with you in a different conversation, I've learned four facts of life which are like death and taxes. Number one, enterprise learning experience absolutely sucks. All of us just hit enter and say, agree at the end. It's just a chip box. And you just hope that there is no test. And if you have a test, somehow get the bare minimum pass. It's not real learning. Second is the what which you were alluding to in your question is our learning

Ram Yalamanchili (35:21.743)

Yeah, that's the checkbox.

Krishna Cheriath (35:37.899)

Experience is very, different today. From back in school to college days, I was always the last minute studier. You can give me three weeks, one month. The studying is going to happen only the previous day night before the test. It is so amazing that that habit is so ingrained in me that even in my work today,

I need the pressure of a deadline to focus and be able to do it. You give me one month to put the PowerPoint deck together, it is going to happen the day before because I can focus. That's what it habits in green. But the learning pattern that I was referring to today is like, when you have in your personal computing space,

The apps that we use on our phone is very intuitive. And if you have a question, you want to just Google it just in time. And then to your point, sometimes you watch a video and see somebody else doing it, and that's where you mimic and you get doing it. That's your learning pattern. So we have to reimagine our learning experience in words to mimic what today's look like. The next most important point, which I would encourage anybody in the audience who is in a position of influence around designing AI or digital fluency programs, is context relevance.

I think unless you are able to connect the dots for your workforce to say, what does AR mean for you in your function and in your role, you're not going to reach the promised land. And final piece that we have to solve for us, information overload. I think each of us working is hit with so many different information hitting us.

that how do you cut through that and get to the bottom line of if I'm working in finance and I have to think about the application of AI to make a meaningful difference in the way I work things, I need to have information that is very focused context rather than delivering a different.

Ram Yalamanchili (37:23.887)

Yeah, like ROI has to be clear to you, right? Like what am I getting out of this for spending whatever time you train on it, right? It's actually a good job, yeah.

Krishna Cheriath (37:28.846)

Yeah. Otherwise, you're always going to get the 20 % who are motivated self-starters who are going to focus on it. And you're going to miss the large part of the audience. And that final bit I'll say to those who are in the position of influence to shape these kind of journeys is when you're selecting a pilot audience, don't go after the quick wins and easy wins of early adopters. That may be the 20%.

I think some of the biggest successes in my career has happened when the implementation of a program, we had a well-balanced set of pilot users between the digitally progressive and the digital skeptics. Because you learn a lot from the digital skeptics and what their skepticism is and how do they work on it. So a few things to consider. But I think that fluency is going to be key. 50 % you owe it to yourself to drive fluency, embrace volatility.

Organizations need to step forward and shape a fluency agenda which is drastically different from historic enterprise learning programs.

Ram Yalamanchili (38:34.649)

Yeah. I am in violent agreement with everything you're saying, Krishna. I couldn't have said it better and I feel like I could probably spend another hour talking to you just about this particular topic, right? And you come from a very, different perspective and also a vantage point right now. You are essentially a buyer, know, slash builder. I'm a builder. So I'm selling, you're buying most likely. So I think my vantage point has been kind of interesting because

First of all, this cycle is all about understanding a type of a technology which has never been really well taught or understood in a broad sense. So an example of that is we are frequently in front of customers now and we're growing, we're expanding. And I'm always somewhat surprised in the early days where the questions the customer is asking, I'm just wondering, you know, you're...

probably not asking the right questions. I can say something and I can convince you that this is the right technology. But if I were in your shoes, I would think about this whole buying pattern very differently. It's that lack of AI fluency because the industry just does not come from that background. Whereas somebody like me, I've spent a good part of two decades in my college. We both went to tech programs. Interestingly, I went to Carnegie Mellon. I don't know if you knew that, but I was an undergrad computer science major there.

Krishna Cheriath (39:41.582)

Mm-hmm.

Krishna Cheriath (39:52.269)

Yeah.

Ram Yalamanchili (40:00.727)

Coming from the domain, I think the understanding is AI is like a software. So does it do X, and Z and you're checking box, you're checkboxing it, right? But the reality is like AI is like actually like hiring a person. It's like you have to ask it a whole bunch of questions. You have to make sure it's like, you know, aligned well. It does the same thing every day it comes to work. It's not having a bad day every three days, you know. So there's like...

Krishna Cheriath (40:14.158)

Mm-hmm.

Ram Yalamanchili (40:26.177)

all sorts of really important, interesting questions and buying patterns, which you have to now think about. And that's buying and then using, right? Same thing. Like you now have to evaluate all these technologies and AI's in a completely different fashion. interestingly, the reason I'm saying I'm in violent agreement with everything you said is I've come to the conclusion, initially, what I used to do is I used to send the people we were talking to early on the early sales.

Krishna Cheriath (40:32.291)

Yep.

Ram Yalamanchili (40:54.351)

Hey, here's like a few papers you can read and you know, somewhere in the literature, I'll just send them somethings or maybe some technical articles about, you this is how you sort of understand these like stochastic or property based machines, which are LLMs. And these are the pitfalls and these are the things you would have to look for. And we are able to solve these problems because we come from a very deep applied AI bench. And you know, we have a large team of researchers who are basically focused on just that, right?

gotten into that. And then, you what I realized is, like you said, people are busy. There's no contextual learning in these type of articles just by reading it. It doesn't relate to what your particular problem is. Yes, an LLM, you know, I can read 20 papers of why LLMs do not have the best self-revaluation capability. Like, you know, you cannot trust an LLM on confidence, right? Like, this is like a very simple concept you and I will like immediately relate to.

But then there are customers will say, based on whatever you did, how do you score confidence? I can say, well, I'll just ask the LLM. LLM is going to be obviously very confident. Every LLM is so confident in its answers, like anything. Because the research, the people who have trained it, including us, we are biasing them towards action. We're biasing them towards agreeability. And we measure that. And we want it to be agreeable. Nobody's going to sit there and use an LLM or an AI, which is fundamentally disagreeing with everything you're saying.

Krishna Cheriath (41:57.614)

Yeah.

Ram Yalamanchili (42:18.371)

So it's kind of like these interesting phenomena that show up and we're sort of like, anyway, my point is that

Krishna Cheriath (42:23.822)

Yeah, think one point I would make Ram, I think it may be interesting for budding entrepreneurs and others who are frustrated with selling into enterprises, especially with the innovative new capability like in AI.

I think I was talking to a founder of a startup, and he was frustrated with, you they said, you don't appreciate the, you're laser focused around the tech leader and the business leader. The business leader is convinced on the business potential, and the tech leader is intrigued by the technology. But you don't realize that there are two nodes in a larger buying puzzle.

To the question of AI fluency, if the legal team in an enterprise is not fluent in understanding and appreciating the nuances of the AI-based capability versus traditional tech buy, if the information security and the third-party risk type of teams are not fluent enough around understanding and appreciating AI versus the others,

You're not going to make the progression in the cell time is not going to be where you need to be. So one of the things that I was encouraging this startup to do was to think about all of those, to use the overused Jerry Maguire line, help me help you, is to say that here, how does, think about it from a legal person who is new to the topic of AI. And what are the kind of questions you can anticipate and how do you answer them?

Think of a security person and others, because that makes us, and you have to, because we're all at different levels of AI fluency. And it is not just the tech leader and the business leader in an organization. It is those surrounding, stereo surround sound that is going to be important as well. So the fluency topic extends there as well.

Ram Yalamanchili (44:16.601)

No, again, I couldn't agree more. in a small way, what we did was we recorded these series of lectures where we basically said, you know what, this is how AI fluency starts in the clinical operations world. It's only focused for Clonops, right? So this is not about general, let me do the math behind a neural network or something like that. And I think it's really just about boiling it down to like small...

incremental steps and saying like, these are the type of things you would think about. These are the areas which, you AI could be great at. And this is how you evaluate it. Ask these questions, right?

Krishna Cheriath (44:51.438)

It was an inspired decision, a decision to do this because I think I've been talking to you from the early days and I didn't see this journey coming around your investment influence. I'm so happy to see it. And I hear about the dramatic increase in the volume of people who are subscribing to it because from that you see that there is an appetite for it. There is an audience for it.

I think what you're making an impact in the industry through that is so profound. giving that to the difficulties in enterprise learning, you're offering one part of the solution by making this accessible in a meaningful way, a contextually relevant way, which is huge.

Ram Yalamanchili (45:36.463)

Yeah, I mean, I think we are trying, but as you said, I really like the way you're talking about the responsibility being shared by both the individual as well as the enterprise, right? think that sort of, I'm very excited about it. I think this is going to be partly how diffusion of AI will start to get into the enterprise. It will be through education and then through action, right? So I would totally agree. Like if you're an entrepreneur in our space, like,

please do focus on education. Your customers don't know what they're buying, nor will they understand what you're trying to sell without educating them. And we over to ourselves, right? Because it's not easy. It's not obvious. So it all makes sense. So that being said, Krishna, I've had a lot of fun talking to you about this. I'm very inspired. Every time we speak, it feels like we're completing sentences for each other.

I hope this was also the same amount of enjoyment for you.

Krishna Cheriath (46:35.298)

Ram, in many ways, the inspiration is mutual. mean, your journey from the start coming as a rank outsider into a space with fundamental questions of why. And I remember some of your why questions at the early days of your tilde journey to the builds you're making. It's amazing. And I've always drawn inspiration from.

the doggedness with which you and Tilda is pursuing this and really supportive of trying to enhance the fluency collectively. think that's big. And I think it yield returns in ways that you can't even predict now.

Ram Yalamanchili (47:13.549)

Absolutely. And I appreciate the support and the leadership and the mentorship. It's great to be in that position with you. So thank you, Krishna. Have a good evening.

Krishna Cheriath (47:24.014)

Great. Thank you so much.

Ram Yalamanchili (47:29.101)

All right.