Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
AI in Regulatory: Agents, Answers, Action!
39 Plays1 day ago
Welcome back to the Ennovation Podcast, where we bring you the latest trends, insights, and expertise in life sciences and regulatory affairs. This time, we're joined by Steve Gens, founder of Gens and Associates, to talk about where AI is really heading in regulatory, beyond the hype cycle.
Steve’s top resources:
- Do you have a business case for AI? Answer a few questions to find out
- Know what to ask vendors: Download the AI Vendor Checklist
- Simplify agency correspondence with AI: Watch AI in action
Recommended
Transcript
Introduction and Guest Background
00:00:00
Speaker
Welcome to the Innovation Podcast, your go-to source for the latest trends, insight, and expertise in life sciences and regulatory affairs, all in one place. At Enove, we're dedicated to empowering life sciences organizations with innovative solutions to navigate the complexities of their industry.
00:00:18
Speaker
I'm your host, Alice, and I've worked in regulatory operations for eight years. My background is in regulatory information management, and I bring my experience working in large pharmaceutical companies to supporting clients here at Enove.
Meet Steve Gens
00:00:30
Speaker
This time on an innovation, I'm joined by Steve Gens, the founder of Gens and Associates. Steve, I'm sure for many of the regulatory world, you need no introduction, but just in case, if you'd like to give us a little bit of background on yourself, that would be lovely. Yeah. Thanks so much, Alice. And, uh, you know, thanks for the invite. You know, it's nice to be on the other side of the podcast as a guest, you know, for a change.
00:00:53
Speaker
For those that don't know me, um spent about 30 or a little bit over 35 years in life sciences. I was an industry guy, Johnson and Johnson or Janssen specifically, and before I went into consulting. And like Alice said, founded the company in 2005. And we do a lot in the regulatory benchmarks um you know all throughout the last couple of decades, do some management consulting, more on the process and organization.
00:01:20
Speaker
and information management layers. So really happy to be here and explore some of the hot topics of the day.
The AI Impact in Life Sciences
00:01:29
Speaker
So today we're going to be discussing, as you say, the hot topics in regulatory and life sciences. And I think top of so many people's agenda is artificial artificial intelligence. It's really been one of those topics that for the last couple of years, I hear from client after client after client when I'm talking to people.
00:01:48
Speaker
How do you think it's really changing the way that we're working in life sciences today? Yeah, that's a really big question. And I think that's an outstanding question. I think there's a a lot of ideas. There's so many pilots, as you know, you know even with you know the end of customer base. So we're really in the experimentation phase. And I think just to start off this conversation and and set context, because we did a lot of thinking about this. And yeah, we had that smaller pulse study. It was 41 companies and Endove was one of 17 software providers about AI and automation and regulatory in the beginning of the year, you know, and it's moving very quickly. But if we take a step back, it's it's one of those things where it fits the hype cycle, you know, and Mars law very well. But
00:02:37
Speaker
As we thought about it deeply this summer, it's really the start of a new era. And I think that's really, really important to start off this conversation that if we think back in the last 30 or 40 years, there's been so many different type of technology plays. But there are some that are so fundamental and that start with the personal computer.
00:02:58
Speaker
So when that came out in the could anybody imagine, you know, kind of cell phones and what we do today? Fast forward to the 90s, that thing called the Internet, you know, you know came out, ah fundamentally ah changing everything and maybe to a lesser degree cloud computing. Right.
00:03:14
Speaker
So here we're in the early days, and that's why I think it's important to frame it as a new era. And why we're saying it's a new era is if you look at the classic, you know, both from taking a new product to market, the whole discovery clinical.
00:03:30
Speaker
and the filing and where the bulk of regulatory work is, is in the lifecycle management. There isn't a spot, you know, where this could not be applied. So a lot of companies, it gets overwhelming.
Adopting AI Strategies
00:03:43
Speaker
Where should we place our bets?
00:03:45
Speaker
And, you know, one of the surveys we did earlier, and it was a nice distribution of the small tier, mid tier and large multinational is all but two companies of the 41. They have a clear AI strategy.
00:03:59
Speaker
you know, set because, you know, from a governance, there's so many areas to play. I mean, the common ones in our world, um and it started out more as an automation conversation is health authority correspondence. You know, there's the ingestion of it. There's the mining of correspondence. And as these things are starting to get sophisticated, where we would bind, you know, the content generation with AI as a QC agent and also translations, when we start putting that together,
00:04:29
Speaker
you know, there's a fundamental improvement. And I think with all the companies that we surveyed and we talk with, you know, the number one business benefit is efficiency. And the outstanding question is, well, how much are we just going to get?
00:04:43
Speaker
Now we're going to get some clues and some answers to that when we, you know, have the results of the survey, our fall survey that we have open right there. Cause what we did ask was what are your perceptions going into a pilot or that business case going in production?
00:04:57
Speaker
And we're looking at, you know, CMC documents, label documents, clinical documents, safety, regulatory intelligence, you know, kind of ingestion. So what was the perceived benefit you hope to get and what are you getting? So so I think um there's not clear data about just how much of a benefit, but I think everybody knows there's a lot of incremental you know improvement. So, yeah, we could talk about this for hours, but that's kind of the high stroke there.
Understanding the AI Hype Cycle
00:05:27
Speaker
And you touched on there that idea of, you know companies are starting to pilot this and this being potentially quite a transformational technology, but you mentioned there as well, the the hype cycle.
00:05:39
Speaker
And I'd love to hear a little bit more about that and and about what you mean by the hype cycle. Yeah. So the hype cycle, I mean, that's a Gartner term, you know, and we've always been believers because we've seen it time and time again. But underneath it all, it's Amaro's law. So what Amaro's law says in the short term, and that's, say, two to three years, things are overhyped. There's a lot of excitement. um But over the longer term, you know, say maybe over 10 years, you know, it's underestimated, you know, the value. So what we see in our data tells us, and even at your conference there in Paris in the fall where we just did that pulse survey of the audience, you know,
00:06:21
Speaker
And it was really clear you know on that data that you know as we get into you know from mid-2026 to mid-2027, that's what we call the implementation point. It's not like, hey, everything's installed. This is wave one.
00:06:38
Speaker
The majority of industry going into production for different types of use case, if it's content generation, AI for reg Intel, um you know, the combination of structured content authoring also, you know, with generative and the health authority, you know, kind of automation. There's a lot that's going to be going in production, mainly for the large multinationals, the mid-tiers. But the small tiers, you know, not that far behind because they like the the bigger companies to figure it out and pay the big money to figure it out.
00:07:09
Speaker
And they they tend to be very quick adopters. so So we're really seeing as we get into next year, um you know, into 2026 and the bulk of 2027, some substantial change.
00:07:22
Speaker
But the important thing in a forum that I was at, you know, just the other day and somebody said something so simple but brilliant. is we need to start looking at AI in phases. So this is the entry phase, right? So entry phase being 26, 27, and then maybe 28, 29, you get into more, you know, advanced automation. So like an example of that is there's the focus on individual document generation, right?
00:07:51
Speaker
There's a tremendous amount of work, the CSR, some of the safety reports, All eyes are on CMC documents now for, you know, kind of M3. But as we start going to two or three years out and as that starts maturing, you know, there are some, you know, of the early innovators is like, well, we know that's going to work.
00:08:09
Speaker
So how far can they go with dossier generation? Right. So I think that's going to be like ah a document generation is kind of a phase one. Dossier generation would be a phase two. And maybe that's, you know, kind of three or four years out for those, you know, that, you know, countries with the ECTD as the structural backbone.
00:08:27
Speaker
Yeah, I think actually the timing of this conversation, I think, is really interesting because, you know, mentioned with with the Mars law that for the first year or or two years or maybe even three years, things are a bit overhyped and we kind of are like, oh, my God, this this technology is going to change.
00:08:42
Speaker
literally everything. And then in the long run, over the 10 year period, that's what we see it be under hyped. And I think we're really starting to hit that tipping point yeah where perhaps some of those early things about like, okay, this is literally going change every single aspect of what we do.
00:09:01
Speaker
We're seeing that those are maybe not going to be true, but then the risk is of course, that then we, we underestimate the potential revolution that we could be seeing here.
00:09:12
Speaker
Yeah.
AI as a Productivity Tool
00:09:13
Speaker
and And one byproduct and it's such top of mind because we've had like three or four sessions just this fall is old fashioned workforce planning. Right. You know, how do we develop and even look at, you know, more of the junior staff that this big question is, well, how far will the technology go? Will we be getting, you know, the ah ROI? What's the impact? Everybody's talking about upscaling, you know, and some organizations are actually having like AI universities where they'll have different levels, like level one Everybody has to be good at prompting because at the end of the day, these are productivity tools. Right. Just like when we all had to learn Word and Excel and PowerPoint. I'm showing my age.
00:09:55
Speaker
Maybe more recently, unfortunately, with the pandemic, we all became experts at either Teams, you know, or Zoom like overnight. So a lot of this is a series of productivity tools that everybody needs to do. But then for more the advanced stuff, and this gets into all the data, data, data, data analytics that we need to start changing job descriptions, you know, so this, this is setting a whole workforce development, you know, traditional career paths for junior people. Are they going to fundamentally change? Cause like you said in the opening, the nature of the regulatory work is still going to do submissions and filings, all that stuff, tasks we do every day. But how we go about that could be very, very different. I think that's such an interesting framing. And i think when you talk about AI as an advanced productivity tool, as opposed to this kind of completely new and completely unknown thing, in some ways that makes me more
00:10:53
Speaker
more able to visualize the kind of impacts that it might have on, on the work that I see people do and on the work that I might've been involved in myself. Um, somehow I feel like when we say AI, it's sometimes something that's quite fluffy and it's quite hard to nail down what that really means. But when we talk about productivity tools, I can picture what that means. I can picture that meaning you know, AI assisting me in finding data in module three documents and pulling that out for entry into a a RIM system or or assisting me with creating those questions for um for training sessions, for example. So, yeah, I think that's so that's a really fascinating framing there.
00:11:39
Speaker
Yeah. And if we take it a step further, you know, because we are one of our sentiment questions in that that pulse survey is, you know, what we are trying to get at it underneath is, is AI going to be fundamentally changing and removing jobs? Because there was a lot of fear in the beginning that, you know, we're all going to sit on the beach and AI is going to do everything for us. You know, and that's just kind of a fantasy. But, you know, the the sentiment question is, you know, is AI an assistant? And like 70 or 80 percent, and this was earlier this year, says, yeah, it's, you know, an assistant. So we think about that.
00:12:14
Speaker
is like AI as a research assistant. you know ai The big use case right now is in you know medical writing. So AI you know as a writing assistant, and then it's not only as the generation of the first version, but the QC, the translations, all that stuff that takes a lot of time and cost.
00:12:32
Speaker
AI as a QC agent, you know there's some concepts that you know In three or four years, are we going have different word plugins that are built on you know AI and the regulatory intelligence?
00:12:45
Speaker
you know So some of the QC that's done by people later on, that intelligence is built into the authoring step on that. And like could you imagine, too, that...
00:12:56
Speaker
You're a regulatory affairs professional, you know, maybe you're in Europe or someplace in you know Asia, PAC, you know, you get that question from the health authority. And this is just a classic example. Well,
00:13:07
Speaker
Has that question been asked by another health authority where your AI agent? So all the discussion now is on agents, you know, the agent, you know, which can consume every health authority correspondence and you consume all the dossiers to say, yes, you know, countries X, Y and z had a similar. This is what the response was. So this is the suggested response.
00:13:27
Speaker
And that could have taken two, three, four weeks to do in the past, which that could just be part of what a new world workflow would look like, you know, as AI as an assistant.
00:13:37
Speaker
Yeah. So it's it's pretty wild times. I mean, I've been around, I've seen a lot of different tech and a lot of different you know, change, but this is, it's just so, like you said, so fundamentally different. And that's, again, this is the start of a new era. And I think that's such a classic example is that, that example of we receive a question from the health authority and previously you'd have to have someone in your team who could say, oh this looks familiar to me.
00:14:03
Speaker
And actually, depending on who you have in your team, that just may not be feasible, you know, but if you can use AI as that that person effectively, where you can say, hey, have we had this question before? How did we respond?
00:14:19
Speaker
What was the response to that response? Was that a successful thing? You know, and you can potentially really reduce your risk and really reduce the delays in turning around those questions. I think that's massively exciting.
00:14:32
Speaker
Oh, it is. And it increases confidence. Like, you know, another classic one with, say, with the kind of quality colleagues and all that is, you know, those boring s SOPs that we are all bound to. Right.
00:14:45
Speaker
So sometimes if there are some, you know, again, to your point from a risk management, like our SOPs consistent, you know, kind of globally. Hey, we hire a boatload of consultants. They're reading them all.
00:14:56
Speaker
you know, having opinions, this is what you have to change. And that could be a six month process versus, okay, it's going to ingest all your SOPs, look for it in discrepancies. So again, it's an assistant, yeah you know, now the downside that could have been more junior people work because that's how you learn the business. You know, you and I had talked about people that start out in publishing and regulatory, that's how they learn the regulatory business. And with a lot of that being automated, it's like how do newer people learn the regulatory business? So like that SOP, you know, it's a classic example that from a risk management standpoint, you know, AI can do that first wave analysis and then the experts, you know, kind of come in and, you know, there's always going to be interpretations and, you know moving things around. But maybe that analysis that takes six months or a year now takes three months or three weeks.
00:15:47
Speaker
Yeah. I do think. I think it's something that, as I say, maybe in that hype cycle, we're shifting now from seeing AI as, oh God, it is going to replace us. It's going to replace these roles to, to seeing it really as augmenting some of these roles and,
00:16:06
Speaker
I particularly can imagine that because my first role in regulatory was data entry. And I think that's something where, you know, I would get information sent to me in emails and documents. I would pull that out and I would put that into a system. And that's exactly the kind of role that I can see AI starting to to do and to do really effectively. But then yeah you're always going to want that human in the loop, aren't you? You're always going to want someone who sits there and goes, yeah, that looks correct. That's not been done incorrectly. And think that's where those junior roles will still exist, hopefully.
00:16:45
Speaker
Well, the junior roles will just be, you know, kind of different. yeah And, you know, at the end of the day, they're productivity tools. It's automation. Because I go back in the beginning of my career and I, uh, you know, one of my favorite jobs, I yeah worked for Water for Crystal when they bought Wedgwood China and they were putting some systems together and I was doing all the distribution finance systems. And and I was fairly new. And, you know, one of the, you know, my colleagues and and finance and accounts receivable, you know, i'm kind of learning how they're using the system. And she was doing something so manual. And I thought like, hell, if I could apply this, attach this to the other system, blah, blah, blah.
00:17:22
Speaker
It basically was saving her three weeks, you know, and for me, it only took about, you know, maybe two hours, you know, to get this done. But I felt bad. I'm not going to say like this was so easy to do. Somebody else should have done it.
00:17:34
Speaker
So I waited a couple of days and went back. But, you know, that was a big deal for that person, you know. So really what I was connecting was kind of that assistant, you know, but it was me actually doing it versus the thing that's crazy about AI.
00:17:49
Speaker
is AI actually has become me looking for those opportunities and putting things, you know, things together. I think that's where the agentic, you know, AI is going the next level. And so we go back to the earlier conversation. Let's all think about AI and phases, you know, or stages, phases or stages, whatever you like better. And we're in stage one.
00:18:10
Speaker
Yeah. And I think, the thing that is so interesting is going to be trying to predict those future developments. Like you said, right at the start, you know, when the internet or when computers first came out, could we have imagined the ways that we use them today?
00:18:27
Speaker
Absolutely not. and And I think it's the same thing with AI is actually the ways in which we use that in five, 10 years time are going to be very different from the ways in which we predict we'll use that today.
00:18:39
Speaker
but actually companies and individuals can still prepare by kind of learning these skills. And it's something, um you know, and I used it on one of my podcasts and geez, I wish I would have written down where, where the originator was. So it wasn't my original kind of ah thought, but it it, it, it was really kind of profound because, you know, somebody said, you know,
00:19:02
Speaker
people are not going to lose their jobs to AI. If they don't use AI, that's the reason they're going to lose their job. But think about it, like if simply if AI is a series of productivity tools that are going to increase, because even in consultancies, you know, the ones that are really embracing, you know, we do a lot of research, like when we get all the data done, um we're going to have our nice reports and our insights, but we're going to throw it in AI to see what it spits out with all that. So, So there is just a lot of opportunity. You know, my brother works at the Everest Group. They do a ton of research, you know, too, and they're really playing around with it.
00:19:42
Speaker
And it's really profound, the ah the quality of the result they're getting. Now, it it it doesn't, it just makes that analyst, you know a lot more effective and a lot deeper insights. And because the assistant can do a lot more research than an individual could. So,
00:19:59
Speaker
So it's really incredible times. And like you said, i I can't even predict, you know, we're pretty good at predicting here, you know, what's going to happen in regulatory, you know, what the regulatory activities with the AI impact, you know, say in 2030 or 2035 is going to be because we're just learning. We're just learning here on stage one.
00:20:19
Speaker
And I think the other thing is that AI is so multifaceted and the different tools that we're seeing are so varied. So we're seeing everything from, as you mentioned, you know, research assistants, writing assistants, QC agents. We're seeing those evolve and develop more and more. And and again, I'm sure that there will be tools, even by the time this podcast comes out, that have come in that we we don't know about and can't predict today.
00:20:49
Speaker
yeah So with that in mind, i'm I'm really interested to hear from the conversations you've been having with various companies. Where are you starting to see companies adding value with AI today?
AI in Regulatory Intelligence
00:21:04
Speaker
And then, you know, how how do you think that will change and evolve over the next few years and and over the longer term? Yeah, so i think adding value, that's kind of an interesting term. And like, where where is the value? Not from a functional, you know, or a department standpoint is, you know, there's companies like yours and Endove that really from a you know, from the transactional system, the system of record, the authoritative source, whatever you want to call it.
00:21:33
Speaker
There's things that you're naturally doing, and you really had a really great, I really liked the ah and no AI roadmap know that you guys presented know in Paris in October.
00:21:44
Speaker
So that's just one example as you know companies like yours and their customers, they're just going to get the benefit as you're discovering and adding things on. just you know It's just automation.
00:21:55
Speaker
So I know you guys are like working really hard on the health authority interactions. That's been like the number one use case for like three to four. And a big focus for us. Yeah. Yeah. Really, really big no focus. I think the other area that I'm really excited about in regulatory and it's been an unmet need is in the regulatory intelligence you know side.
00:22:13
Speaker
And if we break down regulatory intelligence and we do this in like our Reg Intel pulse survey we do every four years, you have the ingestion, right? Which is getting very sophisticated. And then you have the processing. So just think about, you know, with all the biopharmaceutical med tech companies in the world, you know, how many regulatory impact assessments based on regulatory intelligence are done in a given week.
00:22:37
Speaker
So that's where an agent can help with those analysis. And then there's the distribution of that. So you have ingestion analysis and distribution and AI is going to be impacting all those substantially to where you have concepts, like I mentioned earlier, as the authoring is putting together some, you know, tables, listings, it's like, okay, it's going to this country. It has to be, ah you know, in this format, you know, so it's doing QC as you're actually authoring, you know, and that's like downstream process now.
00:23:08
Speaker
So I think yeah it's going to be, you know, very, very substantial. And the thing that I'm most excited about longer term, and I think it's probably three or four years out, is the the automation, especially,
00:23:22
Speaker
There's a growing conversation that started about a year ago, maybe 18 months just you know with concepts um and how far can automation go with M2 and M3 on the dossier and more specifically with M3 with CMC.
00:23:38
Speaker
It's just amazing here. And I know we're recording this at the end of 2025.
00:23:44
Speaker
that in the beginning of 2025 it's all csr protocol that focuses on clinical and it has gone deeply you know and just in the last 12 months to that's really tackle cmc you know tons of complexity maybe labeling but if we can get better structure and tackle the cmc boy how much could we you know, automate, um you know, portion of M3. And that's a really, really big deal, you know, for everybody. So that that kind of innovation, I'm really excited. And there's going to be small stuff along the way, you know, and again, it's in the context of agents as an assistant.
00:24:21
Speaker
And we we already listed probably about seven or eight, you know, kind of examples on this podcast. But I think that the big one for regulatory is the is the ah dossier, you know, kind of automation and really as people are learning things more, applying it to lifecycle management.
00:24:40
Speaker
So that's where the bigger benefit from an operational efficiency is going to come, not so much from the clinic to first filing, you know? um So that that's big. And a conversation for another day, we had it, you know, kind of, you know, hosted too, is the cloud-based ah regulatory spaces, you know, the common DNA nexus and a cumulus So that's the other thing that's fundamentally changing that you can submit know one to many. And that's very early days, too. So I think these two concepts with AI and the cloud-based regulatory spaces are converging over the next two or three years. And it's going to fundamentally change the interaction with the agencies and how things are put together. Just like, you know, you and I can kid around like, oh, yeah, remember the old days, you know, as the ECT came out, like,
00:25:26
Speaker
Before, it used to be all hands on deck for a paper submission. like You just brought everybody in, lined up the table, got your stamps out, and did it. And then you had this thing, ECTD, and the technology where it became a specialty.
00:25:38
Speaker
But you had more structure, and you could start automating. So this is that on steroids where we're going. Well, thank you,
Conclusion and Future Insights
00:25:46
Speaker
Steve. I feel like we've covered an awful lot of ground and we've we've discussed a lot of various topics kind of within that area of ai And I think this is just one that we're going to see evolving further and further. I'm really looking forward to seeing more of your pulse survey results coming out and and seeing more about the impact this is having on companies in in real life. Is there anything you kind of wanted to add before we wrap up?
00:26:11
Speaker
Well, I think just to recap, it is the start of the new era, right? We have to think about it in phases or stages. And we're really excited about the survey that's open. You know, kind of the results will be out around the RSIDM show, you know, kind of in the ah February here in the States, um you know, on that. And it's just where where is the ah the benefit and the, you know, kind of the roadmaps for AI going. So we're going to have some very, very ah detailed information on that. And I think it's going to be telling. So and I think 2026 and 27, it's just going to be a really, really exciting time. There's a lot of learning two, three years ago at RSIDM. Like we all talked about this through PowerPoint, you know, and then two years there's a ago, a couple of companies started experimenting. Last year, everybody was experimenting, right?
00:27:00
Speaker
And I think this year, there's a lot of things in production. These are the results, you know, and that's the thing I love about this community that we both, you know, support. You know, it's a very giving community. There's a lot of sharing. It's not like, hey, in the old days, this is competitive, you know, um that there's so much sharing with companies broadly about what we're learning, you know, with AI and the changes both organizationally process.
00:27:26
Speaker
The shiny object is the tech on AI, right? So it's really, really an interesting journey. So, you know, thank you so much for inviting me and and having me on as a guest. It's been a real pleasure. i hope people will go and check out that Pulse report. And of course, there are always more resources linked in our episode description.
00:27:47
Speaker
And hopefully we'll talk again at some point, Steve, but it's been absolutely brilliant. Thank you so much for joining us. Great. thank Thank you so much, Alice. It's a pleasure. And It's a pleasure being a guest on these two. You know, so the I know the podcasting, you know, it's a lot of fun and it's just, you know, different people chatting and sharing their experiences and knowledge. So, you know, really, really, ah you know, happy to be part of this.
00:28:10
Speaker
Thanks for tuning in to the Innovation Podcast. At Enove, we help over 450 life sciences companies streamline compliance, enhance efficiency, and achieve their regulatory goals our unified platform.
00:28:22
Speaker
Ready to learn more? Visit enove.com or connect with us on LinkedIn to see how we can help your organization succeed.