video

The Future of AI in Life Sciences

Summary Overview – AI will Eat the Electronic Lab Notebook

Kevin Cramer, CTO of Sapio Sciences, discusses the transformative impact of AI on Life Sciences, emphasizing its disruption of knowledge work. AI is already outperforming humans in fields like writing, art, and medical diagnosis. In Life Sciences, AI will enhance productivity, particularly in R&D, by automating tasks traditionally handled by molecular biologists and medicinal chemists. While electronic lab notebooks (ELNs) will remain relevant, AI will significantly reduce their usage. AI will also aid in clinical trials, patient recruitment, and drug manufacturing, though traditional systems like LIMS and clinical data management systems will persist. Sapio Sciences is developing tools to integrate AI with ELNs to accelerate drug discovery.

Impact of AI on Life Sciences

• Kevin Cramer, CTO at Sapio, introduces the topic of AI’s impact on Life Sciences, emphasizing the need for planning and system adjustments.
• AI is disrupting knowledge workers, particularly those who rely on their brains rather than physical labor.
• Examples of AI displacing knowledge workers include writers, artists, medical doctors, lawyers, and programmers.
• Investigate how AI can be used to supplement and enhance the functionality of LIMS and clinical trial management systems, without fully replacing them.
• AI is enhancing productivity in various fields, including medical research, legal research, and programming.

AI in Medical and Legal Professions

• AI is improving patient care and empathy in medical fields, surpassing human doctors in some aspects.
• AI is also making significant strides in legal research and contract review, enhancing efficiency and accuracy.
• The impact of AI on life sciences is expected to be significant, particularly in knowledge-intensive roles like molecular biologists and medicinal chemists.
• AI models are now capable of reasoning at PhD levels and beyond, making them valuable in complex disciplines.

AI’s Role in Drug Discovery and R&D

• AI is expected to disrupt the R&D pipeline, making it more efficient and reducing the need for manual labor.
• AI will enhance the productivity of both good and less capable researchers, creating a leveling effect across the workforce.
• Experts in life sciences will become more like conductors, overseeing AI-driven processes and making them more efficient.
• AI will accelerate drug discovery by handling the heavy lifting, allowing experts to focus on strategic tasks.

AI and Electronic Lab Notebooks (ELNs)

• ELNs are heavily used in exploratory research, but their significance will decrease as AI takes over more tasks.
• AI will handle much of the experimental work, reducing the need for extensive ELN usage.
• While ELNs won’t disappear entirely, their user base may shrink as AI handles more tasks.
• AI will also have a limited impact on laboratory information management systems (LIMS), as bench work still requires human precision.

AI’s Impact on Clinical Trials and Manufacturing

• AI will aid in patient recruitment, data analysis, and clinical study management, but ELNs will still be needed for data aggregation.
• Manufacturing drugs will continue to require systems for environmental monitoring, stability testing, and batch tracking.
• AI will supplement these systems, making them more efficient and user-friendly.
• Sapio is developing tools to integrate AI with ELNs, aiming to accelerate drug discovery and support scientists.

Future of AI in Life Sciences

• AI will significantly speed up various stages of the drug discovery process, potentially making ELNs irrelevant.
• Sapio is working on tools to facilitate in silico design, testing, and assessment of molecules.
• AI will act as an assistant to scientists, helping them do their jobs better, faster, and smarter.
• The discussion on AI’s impact on life sciences is ongoing, with many moving parts to consider.

Video Transcript

Hi. This is Kevin Cramer, CTO at Sapio, and I put together a very brief overview of our thinking around AI and its impact that we’re going to see on Life Sciences. We get a lot of questions about this. I think a lot of people underestimate what’s going to happen, but this does have an impact on your planning and systems that you start thinking about in your ecosystem. And you know, anytime you’re making predictions, there’s always the risk that you’re wrong, but I put my confidence level reasonably high on the predictions we will outline today.

Alright, so first, understand what AI has done to industry in general, even outside of Life Sciences; the experts, you know, if you went back 20 or 30 years, experts said, Oh, it’s the manual laborers who are going to be displaced or disrupted with machine learning, artificial intelligence or robotics, what have you. But it turns out that’s not true at all. It’s the knowledge workers who are being disrupted. You know, people use their brains versus their brains. You see, it turns out to be a lot harder to do physical work with robots, etc., with the precision that’s needed, but to replicate what humans do when we think and have knowledge about things turns out to be much more straightforward than the other. So that’s what’s getting disrupted right now, and there are already a lot of examples of this. I say writers generically. There are all kinds of writers, right? Blog writers, marketing writers, press release writers, book writers, etc, also artists. Now I can, you know, just put a prompt in and get an image. You can even get short videos now, and soon you’ll be able to describe a movie, and it’ll create the whole movie for you, including the script. Bloggers, marketers. AI is writing music, and I suspect in the future, most of the music will be written by AI, not by humans.

Medical doctors have done studies on this. The people who use the systems find them better, more empathetic, and more informative than doctors. And it makes sense. I mean, AI is an Uber patient, and it consumes all the knowledge of medicine that is available publicly. Some of these models are fine-tuned, even to know even more and be disciplined around the medical profession. So they’re already being displaced, and that’s great. We need that because we don’t have enough doctors, lawyers, law, legal research, and contract review, which AI is doing a lot. Now programmers, you know, our programmers at Sapo are way more productive thanks to AI, and that’s just going to continue, you know, and be more and more over the coming years. So we’d be, we’d be naive to think that life sciences aren’t going to be similarly impacted and that it will be the knowledge workers who are going to be affected and disrupted or enhanced, depending on how you want to think about it. And we’ll talk about that in a minute. So this is going to be the next, next year of disruption. I think that you know, doing what a molecular biologist does, or medicinal chemist does, is, is trickier than some of these other things, where you have to memorize in a language model, you know, legal legalese and legal contracts.

You need a certain level of reasoning. You need a specific understanding of microbiology, chemistry, physics, etc. But now that’s already happened, and the newer models now already quite good. They’re at PhD level or better and understand these disciplines important for Life Sciences, and they’re starting to be able to reason. And when you tie this in with tools and safety has done this with our tool of tools, when you start to understand what this means, combining that with an LLM, that can reason this is going to be a massive disruption in life sciences, for sure. There’s no question. And this is not I’m not talking five years out. I’m not even talking two years out. I’m talking now; in the next six to 12 months, you will see radically different things. I think that will have people rethinking where they’re putting their emphasis.

In terms of the R & D pipeline, so basically, people who can do more. And the other interesting side effect of this is that if someone’s a less capable and less good researcher, they will be better. They’re going to be enhanced, if you will, by AI to be more like an expert than they might have been otherwise, and the experts will even be more productive. So it’ll help every. Buddy, it’ll be levels, you know, leveling effect across the workforce. But again, you’ll need fewer people. And I think, as in many other professions, the experts, the people who understand the job, will become more like conductors. They aren’t going to be, you know, necessarily doing the grunt work. The AI will do that for them, and they’re directing traffic, and they’re reviewing and then they’re making another request. Still, the AI will do all the heavy lifting, so it’s just going to make them super efficient, much faster at what they do, and they’ll get to results quicker, and the results will be more meaningful, more viable. So this is all good in terms of accelerating drug discovery. And you know, I’m excited to see you know this this come to be here very shortly.

Alright, so let’s talk about life sciences, specifically, these are very large categories I will talk about here. So don’t you know? I know we didn’t get down into the nuance of each category of of life sciences, but we’re going to start with thinking about R and D as a pipeline and in terms of systems. So, we’re talking about information systems that support these categories of organizational functions. Generally, when you’re in the heavy research end of the, you know, pipeline, it’s, it’s very electronic laboratory notebook, heavy. And that makes sense if you think about it; you’re doing much exploratory work. So you’re doing experiments, right? You’re experimenting a lot to try to, you know, find candidates that you want to take forward. So, it’s very heavy on the ELN front, which makes sense, and less so on the LIMS laboratory information management system. But as you get down the pipeline, you’re going to do more LIMS because you’ll do more bench work. There’ll be a lot more hands on things being done, if you’re scaling up something, for example, on a large molecule, or synthesizing things as saying things there’s there’s more work here. It’s not like LIMS is absent here, but in terms of waiting, it’s a lot more ELN on this side, and a lot more LIMS the further down the the pipeline you get. Now, when you have candidates, and you take them into the clinical study phase, you’re getting the clinical trial management systems, e clinical functions, and clinical data management systems. These are all super important in understanding and, organizing, and planning your clinical study patient recruitment; all the metrics and measurements that are going to be done. This is super important to getting a trial underway and assessing the effectiveness of a trial as quickly as you can. Once you get to market and you start thinking about, you know, you’ve already thought about it. Of course, you’ve done manufacturing even back in the development stage, but now you’re manufacturing drugs that are going to be sold to consumers. So here you care about environmental monitoring, stability testing, Quality Control, seeing your batches and tracking electronic Batch Records, all super essential functions that you need to have, really when you’re taking a drug and selling it.

So, what’s AI’s impact on this entire process? It’s going to be, well, I think, and maybe people will be shocked to think about this or hear this, but AI will largely subsume the ELN market. A lot of the exploratory work is not going to be done by AI. It will be super efficient, super effective, and super fast at working with targets and candidates and getting to something with a lot less experiment, you know, experimental work needing to be done, it’ll be done in silico. And therefore ELNs, they aren’t going to go away. But I could undoubtedly conceive where, in the past, you had 1000 users or on an ELN, now you have 50 or 100 and AI is doing the other 900. It’s doing the heavy lifting for you, so they won’t go away. But I think that’s why I say subsumed mainly. Maybe someday, it’ll totally be subsumed. Probably will. But I think in the near term, near to medium term, I think ELNs will become just less significant in terms of having such broad coverage across the scientists. Because I just know they can do more important things because the AI agent is going to be doing a lot of the work, really, for them.

I think there’ll be more limited impact on LIMS, and that’s because now we’re getting again, to the brains versus Braun question, and now you’re getting to the bench work. You know you need to track that. You have to get on the bench and work. And I think it’s a longer time until we have robots that have the precision to do the work of a lab scientist. So I think for the foreseeable few. Sure lens will still be a critical part of the process and needed in the R and D function. I also think this area will not be replaced by AI either. You still need to consume data from any systems, and AI will aid in things like patient recruitment, you know, analyzing data that you’re getting, but you still need a way to aggregate that data into a single system to look at, to assess how trials proceeding, and also even in the planning stage for your clinical studies.

And here, when you’re actually manufacturing the drug again, systems are going to be critical. I don’t see this changing anytime soon. So I think that these three areas are going to continue to have systems needed just as important as they are today. AI will supplement at Sapio. We’re using AI to help the usability of these systems, to make it easier and more rapid, to build out workflows, to make the user experience using the system more straightforward, using natural language, for example. So we’ll have an impact in that sense. But in terms of needing a system, I think in these cases, it will also be applicable for the next five to 10 years, as far as I can tell, notwithstanding any developments that I can’t anticipate. But I think here’s where there’s going to be the biggest impact of AI is to speed up this part of the process, and we’re doing what we can to make it so that maybe ELN has become irrelevant. So we have our tool of tools, which is in the lab, and we’re looking forward to releasing that early next year, and I think you’ll see some cool stuff coming out of that.

But it’s not just us. There are dozens of places working on open-source or commercial tools that make it easier to design, test, and assess molecules completely in silico. So we want to make access to those tools straightforward. We want to be the Uber kind of integration tool on the ELN side so that we can work with those tools as needed to help get to, you know, a discovery quicker. And we’re excited to, you know, see this future play out and do what we can to support the science and the scientists. And I think the best we can do is to help them do their job better, faster, and smarter. And ai, ai, is going to be kind of the Uber assistant to them in doing their job, so look forward to open discussion on this and other viewpoints. Certainly, it’s ours. It’s a bill, but there are many moving parts here. So it’s, it’s certainly an intriguing discussion to have and think about where we’re headed over the next few years. Thank you.