Ram Yalamanchili (00:07.758)

Okay. Hey, everyone. Today is an interesting episode for us because I thought rather than interviewing more guests, perhaps it's a good time to summarize what we've done this season. And I've got my friend Garo here who I work really closely with is our chief growth officer at Tilda comes with a very deep background in the clinical research space. And for the past couple of years, we've been working really closely in terms of what it means to bring AI into clinical research. And obviously,

working on strategies to launch the product, bring it into customers, and even in many ways educating the market in terms of where we can use AI in clinical trial execution. So I thought this episode we would talk about what has happened over the past eight or so episodes which we have recorded so far and summarize it. And we certainly feel we've learned a lot and maybe share and have a conversation about that. Hey, Garo, how are you?

Gaurav Bhatnagar (01:05.736)

Hey Ram, good to be here.

Ram Yalamanchili (01:08.503)

Yeah. So let's start with maybe the summarization, right? So we've done eight episodes so far. We've covered the various groups of entities or customer profiles which we normally talk to. That would be sponsors, CROs and sites. And I also noticed interestingly, all of our guests have been medical doctors. They're just in different perspectives or working in different perspectives. So...

In some ways, they're experts in their field, they've been in research, and they have multiple vantage points. And I think this summary, at least from my perspective, when I look back, is in research in general, I think everyone kind of more or less point out to the three things, right? There's time, there's quality, and then ultimately there's cost. And I think these are all interrelated. And we've got a very complex system.

an ecosystem right now, which serves its functions in different manners. And we're all working together to sort of advance medicine here. But at the same time, there's quite a lot, I think all of us are looking forward to. I think there's a lot we all earn for. I think AI is one of those breakthroughs which will get us there. So yeah, maybe let's start at that point. What's your sort of like overall impression when...

looking back at these conversations we've had and our learning so far, right, from the industry.

Gaurav Bhatnagar (02:40.274)

Yeah, no, thanks Ram. So I think the two central themes that to me kind of encapsulate everything that we have learned from the different vantage points of our guests so far have been, mean, our system or the lack thereof in the current clinical trials and scale and the challenge of scale. So to kind of break it down and how AI can actually

tremendously either already helping and kind of can really change the paradigm fundamentally. So system to me today, clinical trials either at a resourcing level or, or functional level or at a system level is kind of plagued by a lot of point solutions, like things which, which, which solve a problem, but doesn't really make you look at clinical trial holistically all the way from designing the protocol to submitting the submitting the

And that's where AI, I think, brings computational power. AI removes and amplifies in a very, very significant way. And I think that to me has been either you take the view of vantage point of sites who are struggling with resources or doing trials or recruiting patients or pre-screening patients or from the sponsor's perspective in terms of managing the portfolio of clinical trials.

I think is a central theme which I kind of picked up on from the conversations we had with different stakeholders. The other part is scale. mean, the biggest limitation in clinical trials, which kind of affects very directly the number of molecules we can bring to market or the costs or the timelines is we are limited by a very small group of trained and experienced people. And that's where AI

has the potential of adding infinite capacity, which kind of what I feel is the era of abundance is right around the corner for clinical trials. And I think that's the other part that, you know, can I do a lot more using a lot less? Do I have to do? Do I have to do all of those things? So I think those were the two central themes that came came about and and and resonated with me as we kind of as I was as I listened through all of those different podcasts that we had from different states.

Ram Yalamanchili (05:02.869)

Yeah, and to your point, right, I think when we've gotten our first set of customers onboarded and, you know, partnerships building, I think it was clear to both of us that the challenge is exactly what you're describing. It's not that you can't scale in the current mechanism, it's just really difficult to do it this way. And it presents a really, you know, it's almost like prohibitive to scale in the current methods.

Gaurav Bhatnagar (05:24.638)

Thanks guys.

Ram Yalamanchili (05:31.913)

and we're looking for new methods and the industry needs to move forward, but there are not enough tools and technologies right now which will get us there. And obviously from our perspective, the success of that, yeah, like here's AI related, these are some excellent AI tools which have been designed, trained specifically for clinical research will get you there. And I think that's kind of the resonating factor for us also. I also thought something interesting which came up from

I'll just start with the site side is how much our guests spoke about resourcing. I think that's like a interesting aspect of their business or the way they look at their programs. I think I've gotten a new sense of respect for coordinators, the hard work they do, but also how limited that particular supply is, seemingly from the site owners or the site operators perspective. So...

You know, and I think there's a lot, lot riding on it. How can you improve efficiency? How can you improve training times to get sites competent in certain protocols or certain critical research execution? What are the gaps there and how can we bridge it? And I think there's, there's quite a lot of interesting things which can happen there. I think from an AI perspective, whether it's us or, you know, others in this ecosystem who can help. And there's certainly many others who working on the space, but I'm really excited. think sites ultimately do need.

quite a bit of support and it seems like that world is not too far away at this point from where we're going.

Gaurav Bhatnagar (07:06.47)

No, absolutely. mean, I think that to me was one of the I mean, it is in some ways one of the least talked about things. I mean, if you think about it, the biggest one of the biggest challenges is the ability of train is the is the inflow of trained and experienced study coordinators. I mean, that's just one of the biggest overall and one of the most least spoken about challenges that kind of comes across if you feel the layers of the onion back a little bit and the amplification to that.

critical resource that which AI can is already doing. mean, we are already seeing it across our sites is I think one of the biggest, one of the biggest things. mean, if you could think about it, that all of the manual grunt work that they do and they spend a lot more time with patients is what we are seeing at the sites where we have, where we have already deployed. mean, they screen a lot more patients, their quality of life is a lot better. They are, they are recruiting a lot less. Their data quality is much higher.

And I think that's what technology fundamentally is for, right? To provide that skill and just to give a little bit of a data from today, if you think about it, either depends on which geography you're talking about in clinical trials, but there is about three to 5 % of total investigators who participate in clinical trials. I mean, I think that is too little. the, because clinical trial is fundamentally been super onerous and I think AI removes those barriers. mean,

Nobody really signs up saying that every single protocol you have like 15 systems with passwords to enter and that level of that level of, and I think AI can easily remove some of those burdens. So to me, that's been one of the learnings through the sites we are working with and the guests who spoke so eloquently about that in the previous episode.

Ram Yalamanchili (08:56.403)

Yeah. And I think something you pointed out, right, this platform proliferation, the multiple systems, multiple point solutions. I think I've for some time been following others in this industry, or at least in the startup industry who've been thinking about this problem in a slightly different way, right? I think one of the ones who I've read and I thought has a great argument around this is,

partner contract from Rippling. And he talks about this concept of a compound startup. And, you know, essentially the value of a company or a startup at this point can be a lot higher if you sort of think of your products as integrated products versus point solutions. And in some ways what he's talking about in his example is there are all these HR systems, there's 20 HR systems to onboard people, manage benefits, manage insurance, this and that.

and instead have one platform which talks across the board. And if you can sort of present that solution to customers, that particular solution may not be the best solution in any one point category because you're sort of shooting for breadth versus depth. But there certainly is an inflection point in the markets where that breadth of solution trumps the value in terms of what a single point solution can give in a certain vertical.

I think that's an interesting realization in the clinical trial industry for me anyway. I think there is tremendous value in saying, you know, the value is not just that your one point solution, let's call it your training system can train people and ma'am certifications and manage the overall compliance around it. I think the value is your training system can talk to your EDC system and your EDC system can talk to your TMS system. And at site, maybe it can talk to your source and EMRs and so on and so forth, right?

So having this like really integrated system and having the ability to do it across systems which are not entirely built by you, I think this is the power where you can build AI teammates or AI based systems like we're doing. And it just lends to a much more efficient, much more seamless experience from a client's perspective or from somebody in this ecosystem, right? And I really felt that was the missing piece.

Ram Yalamanchili (11:20.967)

which most of the industry has been talking about, but there haven't been any technological opportunities to do that in a nice way, in a relevant way or a reasonable way. But I think with the many tools and technologies which we have made breakthroughs on in the recent past, just on the AI space, I think those barriers are really quickly kind of going away. What was previously, I would say, the integrators of all these systems had turned out...

ended up being us, like people, right? We were the ones who were managing all these different processes, workflows across multiple systems. And today there's really like very little reason why you can't imagine having an AI teammate or somebody who's your companion or collaborator with you, helping you do some of these tasks and you're just sort of focusing or managing them. So another point I think which I thought was interesting was from the sponsor side, I think.

George McGrath from Opus, Dr. Hamati. I think they were all talking about this concern around how there is a point of no return from a sponsor's perspective in the sense that if you don't find out or if you don't manage the trial in a certain way, you might end up at a point of no return where the program might have a very high risk, which is not building up in it. So in some ways visibility and the ability to change the course of

whatever the trial is, right? I think that amount of control and flexibility is something we've heard of. And some of us, George also talks about how if you have that, that ultimately means that you can run trials in a more efficient, in a much more scaled way, coming back to this scale. So in his case, he's got multiple gene therapies. He wants to manage a large portfolio of trials, but you cannot look at it as an individual trial.

It's more of a portfolio approach and having the insights and the visibility and the nimbleness in terms of how to manage these workflows and complexity and having some kind of a very smart AI companion who can manage some of these complexities along with your team, I think can really like meaningfully change the way they're building their business. So that was like another realization which I think, you know, merits a lot of thought there and I'm actually really excited because of the fact that they were able to raise and

Ram Yalamanchili (13:43.079)

recently gotten some marquee investors onto that platform based on some kind of differentiation, which is not just in the therapies they have, but also like some of the investments they're making on the AI side of things as well. So I do think that's going to continue to happen. What are your thoughts on talking to more of the Seattle's, would say? And you have a deep background there as well, but how are you looking at that or how do you think about it?

Gaurav Bhatnagar (14:12.55)

Yeah, no, I think you're spot on. And particularly from a CRO's perspective, if CROs can partner with sponsors in de-risking their trials in the way like we are talking about, right? What AI can bring in the computational power, the ability to see through end-to-end.

and bring up those risks and allows and brings in that optionality of decision-making, right? Real-time decision-making that has been like, you know, it's been a pipe dream thus far. I mean, I don't think though us as an industry has evolved in the last 10, 15 years at all in taking steps towards that because of that lack of system thinking. And I think that's exactly what George kind of, think, articulates clearly in this thing that how do I really know, get,

transparency of information and data. I there's been a lot of work, I will concede that there's been a lot of work which has been done on the scientific side of clinical trials, but on the operational side of the management of clinical trials about where are your sites, which sites have a greater, what are the blockers at different sites where they need to be starting? What are the challenges around recruitment and how's the quality and how prepared are you for your inspections and how's the data quality that is emerging out of?

I think we have, it's only now that all of those things are coming together and that provides a lot of risk mitigation and capital efficiency, is, think the, which is what we are hearing from, from different CROs at this point in time, because that is that, that provides superpowers in a way like never before. Right. I mean, these, these, this basically makes you like from a sponsor advantage point, it makes you do a lot more trials using, using a lot less.

And you basically are not really necessarily reliant on that much of expansion, human scale capital expansion, which is almost impossible.

Ram Yalamanchili (16:10.511)

Yeah, I think it's an interesting juncture for the CRO industry as well, right? I think it's one of those industries where, as I think about integrators or integrations, I think the CRO industry largely, at least in my view, presents itself as an integrator of operations and workflows, which are relevant to the clinical trial industry. And so because we have scale needs, we have in general needs around time, quality and cost.

I think they're in a unique position where they can deliver this if they're forward on their thinking around adopting some of these frontier technology. But at the same time, I think it's also an innovator dilemma, which I can see with some of them. So it'll be interesting to see where things go. But I'm excited. I think there's a lot of momentum, as we're seeing from the Seattle side as well.

I do think there'll be an interesting sort of like change I would say from the way that industry is today versus let's call it another 12 to 18 months, right? So that seems like it's coming very quickly.

Gaurav Bhatnagar (17:24.82)

Yeah, no, I think that's exactly right. I CRO industry has been, I mean, the history of the CRO industry goes all the way back for like, it's about 50 years now. Like in the early seventies is when it kind of found its feet and it has remained similar. I mean, it kind of serves as, know, it's a specialized experienced resources to the following clinical trials.

And I think this is one of the biggest infection points it's undergoing in the next 12 to 18 months where its model is going to get evolved in a very significant way.

Ram Yalamanchili (17:58.531)

Right. And maybe one thing I also thought of, or at least in all of our conversations we've had with multiple of these different customer profiles, is it is clear that the market sees value in adopting tools or technologies related to AI. That much we know. It's also interesting that how little has been done to get to a production level technology at the moment. I think there are companies like ourselves who are

who've been at it for quite some time and were there. But in general, I think the expectation from customers is really, really high from what they expect AI can do. And that is also a concern I have because I think reality is a prototype to production, there's a very large gap and you sort of need to bridge that gap with technology and investment and focus. But at the same time, I think it's really important that

people understand or the industry understands that, you you really need to run these things in terms of like, where is that production level AI implementation or AI technology, which you can bring in today, right? And I think somewhere along the way, I'm just hoping that the industry does not sort of sell way too forward and sort of like way under deliver on this kind of a situation. But

But I'm also at the same time excited that there will be winners and losers in the way this whole shakes out like any other market disruption, right? So something I've noticed is just even from our guest conversations, the expectations are really high. I think that's one thing I will say about what they feel AI can do and where it can be. And so from our perspective as technologists and others who might be listening who are also in the technology space, I think what we have to really make sure is we deliver, but we don't deliver a prototype.

we deliver a production quality system. And anyone building these kinds of systems understand that nuance between where things can be. So it's one thing to do prototyping and another one to build production systems. So that's another one I think which stuck out.

Gaurav Bhatnagar (19:57.812)

Right.

Gaurav Bhatnagar (20:15.622)

No, I think you're exactly right. mean, there is there is there is that aspect in this industry has been I mean, even in my own previous experiences prior to Tilda, the expectations are high. They are higher at this point in time in some ways and form the tailwinds, which I think the I mean, I would also concede that they are they are good tailwinds. mean, right now, this is probably the first time where the regulators have been on the forefront. I mean, FDA has come out with.

with several use cases and where they are putting together and which comforts lot of sponsors and pharma and the industry that the regulators are kind of coming in front of some of these things, some of these changes. And yeah, but the expectations are high and that's where I think the technologists need to kind of keep that in mind, that what comes out of it has to be, companies like ours, we need to kind of be.

Ram Yalamanchili (21:11.095)

Yeah, absolutely. think that that's actually an interesting take you said, right? I think just recently was at a conference and, you know, one of the audience asked a question to this panel, which was basically regulators as well as a few sponsors on there talking about AI. And I think the question was, what do the regulators think about it? And exactly like what you just said, one of the answers was, well, the regulators are already using it, right? They've already announced that

I think the FDA already has announced ELSA, which is their AI framework to review or AI review system to look at submissions. And so that was an interesting comment in the audience I felt because it was one of those things where people ask this question and then they realize, you know what, there's already an adoption cycle at the regulator site. And they're looking at it from efficiency. They're looking at it from, you know, all the three things we just spoke about, time, quality and cost, right?

So I think that is an interesting tailwind for sure. I've thought about that. And it's one of those things where if you are entirely dependent on a non-AI system to be evaluated by an AI-based system, I think that's an unfair sort of a situation to put yourself in. And so I think the industry will be sort of very quickly moving towards adopting tools which can sort of do

you know, similar things as like the regulators are doing today, right? So there's a great table in there. But at the same time, you know, there's also a discussion in the audience saying, well, we don't know what the FDA has built. We have no transparency in terms of what the system can do, what it's looking at, how they train it, what is the biases there, what is the, you know, all the typical things we would have to publish from a model or AI development sort of a life cycle, right? So we don't have any evals, we don't have system cards on it. So I think

The market is starting to evolve and I'm glad, at least from companies like ourselves, we sort of put that framework front and foremost. But I'm excited. I think into the next couple of quarters over, looks like things are really starting to materialize on the AI side.

Gaurav Bhatnagar (23:27.58)

Yeah, no, interestingly, I mean, I kind of allude to that some of my in some of the conversations, right? We are at a point in time which is probably similar to two times in history when human beings discovered fire before and after were never the same. The other time it was.

I think about 120 years back when we had electricity, right? There used to be the first set of companies which were saying, okay, you're the companies which use electricity and then everything else was electricity. So now we are at a juncture where we are kind of part of that major change and just it will be faster and it will be...

in a way fundamentally change on how we do things. And this industry, I mean, I think it's going to immensely benefit. mean, like think about the complexity of diseases and the targets that we have. mean, from population statistics in 70 years back, the paradigms of randomized clinical trials, they're going to evolve themselves and we'll get smaller and smaller patient populations, more personalized care. And this is how we can maximally reach more patients with the right therapeutics.

It's exciting time, very, very exciting time.

Ram Yalamanchili (24:34.369)

Yeah. Yeah. And I think the other thing which I'm pretty excited about is there are some really like big claims being made by certain individuals who are, know, for all realistic purpose, would say, you know, I would have to give, I would have to take it at face value. So like Demis at Google, for example, right. He's talking about isomorphic essentially running

trials or, sorry, I need to just look up. Is that Isomorphic? What's the name of the company?

Isomorphic labs? Yeah, No, Demis's company is isomorphic labs, right?

Yeah, it is isomorphic. Okay, let me just redo that. Yeah, it is isomorphic. So something I was also thinking about based on your statement, Garub, just now is we are definitely living in an industry or a time in industry where there are some extremely interesting claims being made. For example, Demis, who just won the Nobel Prize and...

now heads Isomorphic Labs, which is one of Google's entities, I suppose. And he's talking about a world where there will be no disease. And he wants to get there, right? And they certainly have the capital. I think they have some of the potentially talent in terms of using AI to do drug discovery. And I'm just curious what happens if

Ram Yalamanchili (26:13.01)

this actually comes together, right? And I mean, we might be many years away from all that, but certainly the reality of that means that we will have to have an infrastructure which will allow this kind of rapid experimentation to happen. And we definitely need better sort of systems than we have today. I don't mean systems technology-wise, I'm just saying like overall the workflows, the way we manage trials and be initiated, that has to happen not only from the industry of clinical research, like, know, CRO sponsors and sites.

Gaurav Bhatnagar (26:26.216)

Exactly.

Ram Yalamanchili (26:42.73)

but also regulators. And I think that momentum is gonna just rapidly accelerate is kind of how I'm seeing it. think there's, we're gonna have breakthroughs after breakthroughs. We've got fundamentally a really strong set of technologies now forming. We have better protein folding models. We have better target discovery models. I think these are all gonna rapidly accelerate in terms of where they'll be next many years. And we also need operational AI, like, you

helping us operate at the ground level and get that data as quickly as possible in terms of the safety profile and efficacy and things like that. So it is definitely an exciting time. I there's some real, really interesting things happening right now.

Gaurav Bhatnagar (27:28.178)

Yeah, no, it truly is. I mean, I often wondered, right, if you're since I started 25, 27 years back in clinical research, right, the burden of clinical, like if you're trained as a physician, if you're trained as a paramedical, if you're trained to to and trained and not just trained, trained and licensed. I mean, there is a very rigorous licensing procedure for medical professionals anywhere in the world. Right. Why is it so much more harder for you to do a clinical trial?

Why is only a fraction of people have the wherewithal to be able to do it? Why is it that difficult? Because everybody's seeing patients, all the physicians trained in the world, only 2 to 3 % physicians actually end up doing clinical trials successfully. It is because the operational burden is so high, because of the lack of some of the systems which can support that, because of the cognitive...

load is too high to be able to remember, okay, this is the ICF, is of this. These are the things. I mean, why can't those things be made more systematized, right? Like how we use an iPhone in a simple way. So I think we are getting to that place in time where that level of democratization of clinical trials, access for patients and physicians to do clinical trials and be part of research and meaningful stuff and bring and be part of bringing

Ram Yalamanchili (28:31.519)

Okay.

Gaurav Bhatnagar (28:51.924)

advanced therapeutics to market is not too far. I strongly believe that.

Ram Yalamanchili (28:55.689)

Yeah, yeah, yeah, definitely. All right, great. Well, thanks for your time. I think this is an interesting recap and sort of hopefully by the time we do more of these summaries in the next months to come, there'll be more breakthroughs and more interesting stuff to share.