Delivering On AI's Potential with Paradigm Health

Speaker 1:
Welcome to TD Cowen Insights, a space that brings leading thinkers together to share insights and ideas shaping the world around us. Join us as we converse with the top minds who are influencing our global sectors.
Charles Rhyee:
Hi, my name is Charles Rhyee, TD Cowen's healthcare technology and distribution analyst, and welcome to the TD Cowen Future Healthcare Podcast. Today's podcast is part of our ongoing series that continues TD Cowen's efforts to bring together thought leaders, innovators, and investors to discuss how the convergence of healthcare, technology, consumerism, and policy is changing the way we look at health, healthcare, and the healthcare system. And in this episode, I'm excited to be joined by Jonathan Hirsch, chief strategy and growth officer at Paradigm Health. Paradigm Health is a clinical research company that aims to make clinical trials more accessible for patients, researchers, and study sponsors through a seamless AI-powered platform, providing solutions for both trial sponsors and research sites. Paradigm Health aims to break down barriers across the clinical trial ecosystem through its real-time clinical trial infrastructure implemented at healthcare provider organizations. Jonathan, thanks for joining us today.
Jonathan Hirsch:
Thanks so much for having me.
Charles Rhyee:
So I just gave a brief description of Paradigm, but maybe if you could expand on that and give us a little bit more history on how Paradigm started and maybe a little bit more of the services that you're providing to the market today.
Jonathan Hirsch:
Absolutely. So the company started with the fundamental recognition that clinical trials are incredibly slow, very expensive, and becoming increasingly more complex and difficult to execute. And I don't think any of that would be news to anyone listening to this podcast. So myself and my co-founder, Kent Thoelke, we came together with Bob Nelson from Arch Venture Partners and basically said we have to build out a platform to solve this core problem. And then from there, we get into the fact that we are seeing an explosion in new drug discovery, in part through AI, but also other technologies as well. So we have way more drug candidates, a clinical trial process that is very slow and not getting any faster anytime soon, and fundamentally we then have a capacity problem to solve. So that's the fundamental basis and premise of the company. So what we're doing to solve that is first we start with the recognition that health provider organizations are the ones who conduct most of the trials for most of the conditions that we all care about, cancer, cardiovascular disease, et cetera.
And we have to empower those health provider organizations with software and tools to make their process more efficient. These are busy clinicians, they spend time seeing patients and clinical trials is, it's not their primary business. It's not what they do on a day-to-day basis. So we provide them with a set of software backed by AI, but software that helps them make their clinical trial process more efficient so that they can do more and more studies. And then for biopharma companies, we provide them with primarily a set of services that are performed by our software to help them design their trials more effectively, to start up their trials more efficiently and choose the right sites, to recruit patients faster, and then to collect data more quickly and turn it into a regulatory submission faster. So that's fundamentally the set of services that we provide. And really you could think about this as bridging two sides of this marketplace, the providers who conduct trials and the biopharma companies who design and sponsor those studies and want the output for their regulatory submissions.
Charles Rhyee:
Yeah, that's interesting when you talk about the bottleneck maybe at the provider. I remember reading some statistics where, and maybe you have an updated number or something like only 5% of providers even participate in clinical research. Is that the biggest challenge that you see today? Because if we think about how few patients who are eligible participate in trials, partly because of how few providers participate in trials as well, where do you think the biggest challenge is right now with clinical trials? Because obviously we all would agree that it takes a long time. These trials probably take longer than they should or they could, but maybe as you see it since you're serving both ends of it, where do you see the biggest bottlenecks today?
Jonathan Hirsch:
Fundamental capacity question that has to be solved, and that's really the biggest area that we focus on as a company. And when you think about capacity, it's not just how many providers are conducting and participating in trials, it's also how many trials can they support and how many patients on those trials can they support? So if you have an individual physician, you have, for example, an oncologist at a community clinic, they're seeing 30, 40, 50 patients a day. I mean, they're running a busy clinic, and if you're asking them to do a trial, that's a lot of activation energy. And if we can then say to that physician, we can enable you to do 10 trials and put 50 patients on those studies with the same time and the same effort as you doing one trial with two patients on that study, that opens up a tremendous amount of capacity in the industry.
So what we see is in terms of problems, the first is the specific burden that the trial operations place on the health provider. And I think this is probably something that folks on the sponsor side don't quite fully appreciate. The trial is incredibly burdensome. There's a lot of activity required, there's a lot of busy work, paperwork. So the first problem that we try to solve is the burden on that individual physician to conduct the trial. And we have a bunch of software and solutions that help take a lot of the busy work off of their plate. The second part of this is then the patient identification and recruitment. So you can take all the burden, you can, for example, redesign the protocol, which we do to make the protocol less burdensome. You can build AI software that helps reduce the busy work, but then you actually have the patient recruitment problem.
So we actually address that in a few ways. One is we built out a bunch of LLM-based software that does a lot of the work of finding patients, reading the patient's medical record, assessing eligibility, but then we go beyond that and we've actually built out agentic based care navigation workflows to help navigate the patient into the study, surface them at the right time, close care gaps. So we do quite a bit of work on that side, but we also address some of the structural issues around how you route patients and refer patients within organizations to trials. So what we're specifically doing there is we're bringing the clinical trial to the health provider organization that already has the patients rather than referring the patients out of the provider organization to a different clinical trial site.
Charles Rhyee:
Can you talk about how you guys do that?
Jonathan Hirsch:
Yeah, absolutely. So what it starts with is the relationship on the biopharma side. So we work with the biopharma company and we help the biopharma company first with protocol optimizations. So we're typically involved upfront in helping optimize the actual protocol. Then we are able to take that protocol, analyze the patient population across all the providers that we work with, but then we can very specifically see which physicians actually are seeing the patients versus who the pharma company might want as the PI. And we can also see geographically where those patients sit within that provider organization because a lot of these providers are big geographically distributed. So we then go with all of that intelligence and then also a little bit more than that, which is are there competing trials that might compete for the same population? So we go with all that intelligence and we sit down with our colleagues at the pharma company and we develop a comprehensive recruitment and site activation strategy with them.
And we sit with them and we say, "If you put the trial at this community oncology provider, we can accrue more patients more quickly. Maybe it's not the PI that you want, but this is where you're actually going to find patients." Or we say, "We can set the trial with the PI that you want and we'll implement a strategy to have the patients from their colleague physician referred over to them, but within the same institution so that that health provider is not losing patients." So the important part about all this is that first, it relies on a comprehensive agentic platform. So the platform is actually doing a lot of this work, not humans. The second part is that this is highly consultative. So we are understanding the business dynamics of the individual provider. We're understanding the business and scientific objectives of the pharmaceutical company, and we are figuring out how we actually accomplish both parties' needs within the constraints that are presented to us. So it's a highly consultative process, but one that is incredibly scalable because a lot of the work is actually automated by the software that we've already built out.
Charles Rhyee:
Yeah. What I think was interesting how you talked about it there, it's two pieces. There's the technology piece and obviously everyone's all excited about AI and that's the entire discussion, not just in healthcare, but across broadly in the markets. On the other side, you're talking about also having to have deep domain knowledge about the process, about workflow, about providers, et cetera. And that raises the question, I think, but maybe talk through, because everyone's using the words AI in all their discussions. So maybe a little bit, how do you differentiate the technology and the AI platform that you've built?
And so how do you differentiate that? Because I'm sure a lot of competitors are coming to same biopharma sponsors and say, "Hey, look, we are using Agentic AI, et cetera, and we can do all this stuff." So I think there's a lot of hype out there. Maybe help us understand. So maybe two questions. First is how you guys differentiate yourself in the market. And then secondly, maybe for the listeners, when people hear the words AI, and maybe in specifically let's say in clinical trials, what would you say is real versus hype? And maybe help our listeners here understand what to take seriously, what maybe you can ignore.
Jonathan Hirsch:
Yeah, absolutely. So for us, when we talk with pharmaceutical companies or biotech companies or health provider organizations, when we talk about differentiation, I'm a big fan that you differentiate based on the actual outcome. So all the software, all the services in the world, it doesn't really mean anything unless you actually have the outcomes that matter. And for us, we have demonstrable outcomes out in the marketplace. For example, with one top five biopharma company, we've been able to increase their per patient per site per month accrual rates across the portfolio of phase one through three studies by 36%, that median across their studies. Some studies were up to 395% faster. So this is a very substantial time savings, cost savings, efficiency gain for this particular sponsor, but that's replicated across pretty much every sponsor that we work with. The software reduces latency for collecting data for the clinical trial from months or sometimes years for study data collection down to days.
So 1.9-day average latency for data collection, reduces data errors, substantially 91% reduction in data errors from human transcription to the software. So we differentiate based on actual demonstrable outcomes that folks care about. And I feel that that's very important that you don't just provide software and you expect that someone is able to generate value. You have to go all the way to actually providing the outcome for someone. In terms of AI hype versus reality, so you look at the outcomes that I described and you would say, "Okay, well, how do you generate that?" For us, we generate that in a scalable manner by having qualified humans teaching our AI software how to do the job more efficiently and more effectively. So anyone today can get ChatGPT or any of your favorite AI tools. You can plug in some data and start using it. And you probably notice anecdotally as you use these services that they actually get better over time as you use them and as you expose it to more and more knowledge and context about you and the work that you're doing.
Well, we've actually taken this to the max. So we've built our own internal AI stack, not a foundation model. We use many foundation models from various different providers and we actually switch them all the time to get better effectiveness, cost efficiency. But on top of the foundation model, we've built our own AI stack that is specifically tuned for clinical trial use cases. So we've actually taught the software about what patient recruitment is. We've taught our models about what it means to find a ECOG score in a physician's note and to pull that out and extract that for a case report form. We've taught the software how a research coordinator works, how a data manager works, what the PI's workflows are. All of that is actually stored in our agentic layer in specifically a memory and clinical training layer that we've built internally so our software understands what the workflows are. And then the thing that we have the advantage on is we actually have people who use our system.
So we have research coordinators, managers, PIs who use our software, and by using our software, they actually create their own agents in our system. So they train the software to do work on their behalf. So a PI is training the software to review data on their behalf and the software gets better and better over time. So in terms of how I think about very specific use cases today, so the clearest use case where AI excels is in taking all of the historical patient medical record data and reviewing it to assess eligibility for a clinical trial. That's the obvious use case where AI excels today. And the reason why is there's a ton of data, it's unstructured, it's messy, and you can use an LLM to help interpret that information. It is very difficult because you actually have to train the LLM to understand the medical concepts and really to iterate and refine on what you're looking for.
So it's not that these things work out of the box, but the LLM is a very good tool to help you do that. Things that are, I would say, more hype at this point are the AI agent is just going to design your endpoints for the trial without a lot of input. That's hype at this point. I think there's a direction of travel where that does become reality. For example, internally, we've already built out a capability to take an unfinished protocol, clinical trial protocol, and we run it through our Agentic software and our software actually suggests the changes that need to occur to the protocol to make the protocol less burdensome for the physician and the patient. So we've already built that and that runs internally and it's part of our trial design service that we're launching, but it's not something that can generate the scientific endpoints of the study yet. I think that we collectively are probably a few years from that.
Charles Rhyee:
Yeah. And you mentioned that you're demonstrating outcomes and that's the differentiation. Curious, maybe talk a little bit how your platform is used by biopharma sponsors in particular, because you're walking into an environment where they've already have significant investments in technology and most of them have a tech stack of some form or another. They're partnered with dozens of vendors on average to partner with you. Is this something where they have to, is it they can integrate you into what they have? Talk about how you're able to work within your customer's environments.
Jonathan Hirsch:
I think that achieving the systemic change that we want to see is a big enough undertaking that we then have to be very flexible when it comes to how we engage with pharma companies to achieve that, and you kind of have to meet every pharma company where they are. So for us, we have three primary things or solutions that we sell into pharmaceutical companies. The first is what we call plan. So it's our trial design and feasibility offering. The second is recruit, so that's our patient identification, recruitment flows, recruitment reporting and analysis. And then the final thing is what we call conduct, and that's all of our study data collection, data verification, monitoring, and submission of the data to the sponsor and then to the FDA. So we sell those three things. We also sell a full end-to-end solution that we call Spire, which is our ability to basically do the whole entire study.
They just outsource it to us and we run the whole trial for them. So we are then quite flexible in terms of how we engage. What we typically see is that every biopharma company, regardless as to what they built in-house, regardless as to whether they outsource some components to CROs, for example, every pharma company has trouble with designing a good trial that meets the right patient population and reduces burden. Recruiting patients, everybody always wants that faster, and collecting data in a less burdensome, high quality manner. Everyone wants that faster, lower cost. And for big biopharma companies, how we typically engage with them is either they come to us because they have a very specific problem, so they want to solve a recruitment problem for a set of trials, or they come to us because they want to do a broader transformation and they're engaging with us across all of our efforts.
What invariably happens, because this is how all biopharma companies adopt, you start with one thing on one study, one solution on one study, you prove it, you then do two solutions or one solution for two studies, you prove it, you then get five studies, you prove it, and then you get 10 studies, you prove it, and then you are a preferred vendor. And that's what we've done, and it's incredibly difficult. We've come up that evolution with a substantial number of the top 20 or 25 biopharma companies, primarily focused in oncology, phase one through three studies. So we've come up that journey with a lot of them, and we're incredibly pleased to be partnered with a number of the large companies as a preferred vendor to them.
But I would say what's consistent for all the top 20 is first, they want to change their operating model. They're not going to do it overnight, but they're looking to companies like ours to help them effectuate that change. And they want to be more site friendly is a big thing for all these companies right now because in the United States, clinical trials are more and more difficult, sites turn down studies, there are tons of trials competing for their attention. So if you are a biopharma company and you have a hundred oncology starts per year, you're actually not just competing with other pharma companies, you're also competing with yourself. And all of these pharma companies want to be super site friendly. They want to reduce burden on sites.
They want to use technology from Paradigm Health to help them do that, and that allows them to more effectively compete with other pharma companies for time and attention, but it also means that maybe someone's going to prioritize their portfolio rather than someone else's given how many studies they have. The final thing is I think that there's a broad recognition, and this has been what's changed over the past few years, there's a broad recognition that we're fundamentally in a very interesting geopolitical moment where a lot of the trial volume has been moving abroad and frankly has been moving to China. And we're now at a place where a lot of pharmaceutical companies are understanding that they have to do things to invest in building United States site capacity, otherwise it's a fundamental long-term issue. So we've seen that very strongly from a lot of the leading pharma companies that not only are they investing in manufacturing in the US, but they're also investing in the clinical trial ecosystem in the US.
Charles Rhyee:
Some stuff to unpack there. I guess the first question I'd have is, at the beginning you said our goal is to meet customers where they are. So I guess the question is where are customers then? Where is pharma today? And obviously I think we've all seen them, a lot of restructurings have happened overall, a lot of discussions of how their models in terms of their R&D priorities and their portfolios, all that stuff has been happening. This focus on let's say technology and AI in particular, I'm sure it's not that AI is new, but this hyper focus on it maybe let's say in the last year or so. Where are biopharma companies in terms of their strategies as it relates to AI? And maybe if you split the market between, let's say, top 25 pharma versus maybe mid-size pharma companies versus biotech and emerging biotech, how are these companies all, where are they in their journeys and how do you fit in with those?
Jonathan Hirsch:
The vast majority of the top 25 biopharma companies are moving fairly quickly for organizations of their size. Certainly a global top five biopharma company is never going to be the most nimble organization, but given their size, they're moving incredibly rapidly and it's partially because of the pressures that you see in the market. So if we're going to either moderate drug costs or we're going to be on a trajectory to try to keep things stable, something has to give. And one of the things that has to give is you have to reduce the time of clinical trials. It's not necessarily that the clinical trial cost itself is the main contributor, but it's the overall financial calculation, which primarily is driven by the risk and time aspects of this. So the part that we can control and we all can control is clinical trial time.
So if we can shrink the trial time by 10, 20, 30% from IND to NDA, you're having a substantial impact on the economic equation for investment in a new drug program. So the pharma companies are looking at this and saying, "Wait a second, if I can actually spend a little bit more money and do a little bit more change now to lean into AI technology that helps me shrink that time, it's an investment that is worth it." And frankly, it's an experiment that's worth doing. The experiment of AI enabled versus traditional, that is the specific experiment that most pharma companies are doing internally. We are part of, I would say, most of those experiments, many to most of those experiments right now. So when we're working with a biopharma company, they're never saying it's all or nothing at the outset. They're looking at a paradigm technology enabled site and paradigm influenced trial design, paradigm influenced site selection, study data collection.
They're looking at that versus the comparable traditional way of doing it. They're concluding that we are faster and at the same or higher quality, and then they're deciding that this is a way to go. So I would say most of the top 25 biopharma companies have active programs to lean in and embrace this. It's moving faster than I would have anticipated. Part of that is being driven by pressures politically. So for example, the FDA's Real-Time Clinical Trials Initiative and some of their other efforts that we'll get into, which help basically remove some of the friction of what will the regulator think. So these efforts are moving forward fairly quickly. If you look at, let's go the other direction, if you look at small biotech, I would say that small biotech is embracing AI in preclinical, but they're a bit risk-adverse in terms of embracing this in their actual clinical trials.
And I think a lot of this stems from the fact that if you're a small biotech company, you've raised some venture money, you're funded to do one trial, you're probably going to look at this and say, "I'm not going to quote unquote risk." I wouldn't perceive it as risk, but from their side, they probably say, "I'm not going to risk my one trial and one asset on this new approach. I'm going to do the standard approach even if it's slower and more expensive." And I think that for small biotechs, they'll probably be the later adopters of this, ironically, because of that risk profile. Where things get very interesting is those mid-size biotechs, some of them are public, those are the companies that are actually able to move incredibly quickly. So we actually see a lot of very rapid movement from that segment because they want to embrace technology. They think it's going to help them move faster.
They don't have the time to waste. They're in a super competitive market. They have portfolios so they can experiment with some of their clinical trials. And they're more nimble organizations. They don't yet have all the big bureaucracy of any large company, let alone a biopharma company. So that's how we see the market playing out. But I do just want to emphasize that it is really surprising in some ways and heartening in a lot of ways to see how quickly the large pharma companies are moving almost uniformly. There are a few who don't have big R&D pipelines and are facing patent cliffs. They have different considerations where they're using AI for more staff augmentation, things like that. But most of the companies are moving incredibly quickly.
Charles Rhyee:
Yeah. And you mentioned it, that definitely seems like some of that has driven top down from FDA. I think FDA really talked about implementing more of a real-time clinical trials, and I think they had an RFI for a pilot. I don't think we've gotten the outcomes of that yet. Maybe you can update us there. I think they did an RFI back earlier this year. I know that you guys submitted this summer in response to the RFI. Maybe talk a little bit more about what the FDA is looking at, how you guys think about participating in it, and what that means, you think, for clinical trials?
Jonathan Hirsch:
Yeah, absolutely. So the FDA approached us in December, late December of 2025 actually through Amgen, interestingly enough. So they approached us with what they thought was a very audacious idea, which is could we use AI driven software to help collect clinical trial data and then to analyze the data to turn it into what they call signals and then send those signals to the FDA to enable much more rapid review of clinical trial information. And we said, "That's amazing. We've actually built out most of what you're asking for. We just don't have the submission of the signals to the FDA part yet because we haven't talked to you." So we had a real mind meld on that. We signed a collaboration agreement with the FDA within a few days of that, and we then did a very quick initial proof of concept within about a month of that. So that's in January 2026, so this year.
And then we started building out with the FDA what became the Real-Time Clinical Trial Initiative. So effectively what that is, is a program from the FDA where the FDA accepts a trial into this RTCT program. That means that signals from the trial that are predetermined by the biopharma company, the FDA and us at Paradigm Health, those signals are codified into our AI software that operates within the health provider organization. And as the clinical trial is occurring and our software is collecting that data, we are running those signals and then transmitting those signals to the pharma sponsor in real time for them to review. And then for certain signals, we send those to the FDA directly after that. And then other signals have time-based triggers for when they're sent to the FDA. So we got that program up and running on the first trial, which is AstraZeneca's phase two TRAVERSE study. So that is live and running at MD Anderson and New Penn and sending signals to AstraZeneca and the FDA on a regular basis.
And then we also initiated another study which was different because we actually designed the study specifically to be one of the pilots for this initiative. So it's with Amgen, it's a phase 1b called the STREAM-SCLC study. So we have these two trials and they're the first two, what we call proof of concept studies with the FDA. So those studies are running and part of this FDA initiative. And the FDA publicly announced this at the end of April of this year. And what we're now in is the scale-up phase. So the FDA put out an RFI, which is leading into a formal request to specific sponsors to include specific clinical trials in the scale-up phase. And we'll probably do somewhere around 10 to 15 studies in this next scale-up phase is what it's looking like. And really the idea behind this is can we shrink the regulatory decision time for moving from phase one to two and two to three? That's the specific focus of the effort right now.
So we're not looking at the final approval, we're not looking at the final NDA or BLA. What we're looking at is the move from phase one to two and two to three because those are key decision points that we believe collectively can be shrunk fairly dramatically in terms of time. Eventually, what we're going to do, what we hope to do is move this to looking at phase three trial submissions. So this is one part of the FDA's overall initiative. So they announced several other related initiatives. So for example, they have an effort to reduce the time for submission and review of INDs. That's another part of this. And all of this is about how do we make things easier and faster for companies developing new drugs, especially so that those clinical trials happen in the United States. And that has removed a lot of the, or is in the process of removing a lot of the friction for pharmaceutical companies.
Charles Rhyee:
Maybe one follow-up there is with FDA then receiving data in real time, I mean, is the idea to allow ongoing discussions with the FDA as data comes into early termination to move forward? Is that the idea that like, "Hey, we've seen enough, we can cut to short," or does this allow FDA to intervene and say, "Hey, look, I'm sure you've seen this, but we see something maybe concerning. Let's have a discussion."
Jonathan Hirsch:
Yeah, exactly. Now, the important thing to point out is the FDA is not getting any of the raw data, so they're not receiving raw patient information, anything like that. The FDA is receiving what they would otherwise receive, which are processed, adjudicated signals and data sets, et cetera. So it's just giving the FDA what they already get, but giving it faster. And I think there's a lot of common misconception around how the FDA does their reviews and what data they ask for and all of that. Today, when a pharma company or any sponsor is preparing a submission, I mean, it takes them months and months and months and months to put together data packages.
It's a very long time period. And then when the FDA gets it, they then have to pull everything apart and they have to reanalyze and readjudicate it. So we're introducing at least a year, probably two years into an otherwise long process. So what this is designed to do, and I think this is very important, upfront in this process, we have the FDA, the sponsor, and us sit down and say, the FDA says, "Here's what we actually want to see and what we want to review. Here are the data signals. Here's what we're looking for." And then the sponsor dialogues and says, "Well, here's what we think is important, what we want to analyze." And there's a discussion around it to come to an agreed upon perspective.
And then we take that and we codify that as automated review rules into the system. So we're taking this whole two-year back and forth odyssey and we're just shrinking it down and codifying it so that the FDA gets what they want faster with less burden. And there's still, of course, the ability to then go back and argue and debate and refine over time. But the point is that we're taking out a very long process of basically collecting data, converting it, sending it, unbundling it, reanalyzing it. We're just removing a lot of that friction. And again, it sounds simple and conceptually it is. It's very difficult to implement, but it's something where it's just a common sense way to take some time out of the process. Is this going to magically take a trial process from 10 years down to five? No, but if we can take 18 months out of a process, that's still pretty good.
Charles Rhyee:
Yeah. And at the end of the day, that is what the potential of technology and particularly the advancements we've seen in AI can lead us to. And it's intuitively, yeah, of course, why should this take as long as it does? But when you break it down, you can see from a regulatory standpoint what you're required to present to agencies. And to me, it sounds like this is why technology is important here and the opportunities. And so maybe really as a last question then, as you think about what this looks like maybe over the next five, 10 years, what do you think the clinical trial of the future looks like when all of this is incorporated? And you talked about you see even a road towards trial design as well. Maybe talk through how you and Paradigm sees the trial of the future, what that looks like.
Jonathan Hirsch:
If you think about what happens today, someone's writing a trial from memory, they're writing a protocol into a Word document, they're sending the Word document out for review, the Word document goes to the site, the site finger in the air says, "Oh, I think I have enough patients and capability for this." And then every single patient, they're doing manual chart review. And when they collect data for the study, they're writing it on paper and then retranscribing it into a system or they're typing it into their EMR and then swivel chair medicine into a data capture system. And then the pharma company is either flying one of their own or they're paying a CRO to fly a human being to that provider organization to look at how data was manually retranscribed from one system to another and monitor that. That's the current state of affairs as the baseline. The future state that we imagine is and that we're trying to achieve is that you have a nationally deployed platform that is intelligent about all of the patient workflows and patient process through their care.
It knows about the doctors and what their interests and capabilities are. It knows about what the capabilities are of the individual hospitals and clinics in which they practice. And all of that information can then be leveraged and used in an automated manner to create a clinical trial design that works for the care settings that that drug is going to be used in. And I think that is a key important point. If the drug doesn't work for actual patients, what's the point? So the first part that we imagine is that a trial is designed through mostly technology that actually reflects what real patients experience. And then from there, you can have AI agents reading the patient's chart and adjudicating eligibility, taking the work off of the physician's plate so the physician can actually focus on having a conversation with the patient and providing care. And the conversation is a conversation about here's a clinical trial. This trial is for this drug. We don't know if the drug is going to work, but you are going to get the highest quality care if you participate in this trial.
And we can then focus on that conversation rather than all of the busy work behind the scenes. And then as the patient's on the study, the physician, the PI, isn't doing any of the manual labor that they do today to run the logistics of the trial, that we have AI agents operating behind the scenes, documenting all of the data about that patient for the trial. When the physician is doing their visit, a voice agent is listening and recording all of the clinical trial elements and automatically putting that into a study database. We have AI agents analyzing and verifying all of the data, and we have an AI agent doing all the medical review of this and the submission to the agency. So what we're trying to achieve is a fully agentic workflow for a trial that is embedded within a provider organization so that we can then create more capacity for every biopharma company so that we can realistically 5X the capacity of the United States to do trials. That's what we're trying to achieve.
Charles Rhyee:
Well, we'll look forward to that. I think that's a very, it sounds an achievable goal actually over time. Certainly exciting to see how things progress. So Jonathan, really appreciate all the time you've given us today and look forward to seeing the progress that you guys make and I look forward to hearing for updates soon.
Jonathan Hirsch:
Thanks so much for having me. It was a pleasure.
Charles Rhyee:
Great, thank you. And thank you everyone for joining us today.
Speaker 1:
Thanks for joining us. Stay tuned for the next episode of TD Cowen Insights.