Elisabeth Schwartz [00:00:00]:
That's the lift with these AI systems. You've got to get the data in there and usable and have it be trustworthy. You've got to trust that the engine is scraping and regurgitating information correctly. It could be wrong if it doesn't interpret it correctly. And that's it. It's the data. It's understanding the data. It's making sure that the model understands the question and then responds accordingly.
Elisabeth Schwartz [00:00:20]:
That's the testing you have to.
Steve Swan [00:00:26]:
Hello. Welcome to Biotech Bytes. I'm your host, Steve Swan. This is where we talk to CIOs within the biotechnology field about their thoughts and feelings around technology that's currently affecting our industry. And today I have the honor of having Elizabeth Schwartz from Fathom with me. Welcome.
Elisabeth Schwartz [00:00:47]:
Hello. How are you?
Steve Swan [00:00:49]:
I'm doing well, thanks. Thanks for being.
Elisabeth Schwartz [00:00:51]:
Pleasure.
Steve Swan [00:00:53]:
Thanks. And so usually what I like to do when I start out, Elizabeth, is just kind know, start real basic. And it seems in the recent past a lot of folks have been kind of going after the OR talking about the low hanging fruit, which is AI. Everybody's talking about AI. Everybody's hearing about AI. Everybody's reading about AI, not only us in the technology field, but the folks that are in the business and such. And just wondering where your thoughts are there, where your feelings are there, where you are on that, where you are in AI right now.
Elisabeth Schwartz [00:01:27]:
I'm a big fan of AI for certain business processes, and we do actually have one group that has purchased a system and we're in the midst of implementing it now. What I find is it's really the subject matter expert lift from an IT perspective. It doesn't take much time and we're not involved in the coding. It's all about the business user getting the artificial intelligence to respond with answers that work for.
Steve Swan [00:01:56]:
And do they have many use cases? Again, don't need specifics, but many use cases right now for AI that it's live for.
Elisabeth Schwartz [00:02:07]:
Not yet, but we're close. There's really two types of AI systems that I'm seeing. The first is a completely self contained system where you would load policies or documents that have information on them, and the AI system scrapes the data, puts it in the system ready to use to answer questions, and the AI integrates with your Microsoft Outlook experience. So you can type into your team's window what's the meal limit in Chicago for dinner? And get an answer back on what the policy regurgitates. So that's one type. The second is one where it does have connections to the outside world to look for research. The issue with that, there are two issues. One is you need to make sure that the data you enter is never put live on the Internet.
Elisabeth Schwartz [00:02:58]:
Because there are cases, like, there was a case with Samsung where somebody put a presentation into the chat GBT, and it got all over the Internet. It was a private presentation. They didn't realize the information would escape.
Steve Swan [00:03:10]:
I didn't hear about that.
Elisabeth Schwartz [00:03:11]:
That's crazy.
Elisabeth Schwartz [00:03:12]:
I'll send that to you afterward. I know, right? It is crazy. So the one rule we've given our business users is do not put sensitive information into chat GBT. If you want to get a system that's self contained or protects the information, we're all on board to help. But that's the one non negotiable piece.
Steve Swan [00:03:30]:
Now, someone told me with Chat GPT, for it to actually remember your data, you've got to be very deliberate and toggle it on. Save my data to learn to tweak the model, or you have to toggle it on to say, when I give you a thumbs up for what you return to me, save my data. Otherwise it forgets your data. It's in your terms of service or something along those lines. Right. And they were trying to tell me it was a pretty heavy lift to do that. And it doesn't sound like it is. It sounds like it's pretty easy, right?
Elisabeth Schwartz [00:04:03]:
To me. Yeah, absolutely. Yeah. I'm sort of surprised about that. There might have been one of the systems that this particular person was looking at. I have not seen that. But then again, we're, like third level users of GPT. They make GPT.
Elisabeth Schwartz [00:04:20]:
You have systems that harness that coding and create a system for the end users to use. So we're getting it after both of those pieces. So there might have been a step along the way that might be more complicated, but we're not seeing it where we're looping in.
Elisabeth Schwartz [00:04:34]:
Okay.
Steve Swan [00:04:34]:
And now your organization. I mean, we can talk about the tales of two worlds, right? You've been at big companies, you've been at small companies, right? Big companies have a lot of resources to dump into things like that. Smaller organizations, not as much, right. Because, listen, everybody's got a finite budget for everything, right? But the bigger companies have more to dump into these things, and some of them are actually, they have an innovation budget, or I don't think they have an AI budget, at least not yet. Right. That might be coming. But have you found that because your company, how many folks are in your company, total?
Elisabeth Schwartz [00:05:06]:
We're getting closer to 500.
Steve Swan [00:05:09]:
Yeah. So relatively small compared to some of the others. Right. So sometimes some of these systems, sometimes you can't do everything you need to or want to with them. Right. And then you get caught into a situation where some of the bigger players, like I said, can run with it. Right. So they got a lot of resources to dump into these things.
Steve Swan [00:05:28]:
So would you give any advice to other folks in smaller companies like your own, where you would say, hey, listen, if I knew this a few years ago, I would have done this differently or something like that coming from, because you were in larger companies before and you came here. Right, to a smaller place, is there something that you would think about differently that you know today that you didn't know then or. Not really, not much.
Elisabeth Schwartz [00:05:54]:
I find that being a smaller company, it's more efficient for our company to use vendors rather than get specific subject matter experts. I think that would be the one difference. If you're at a large company, you probably need to hire more people because you have a need for it, whereas we may only need 10 hours a month of an oracle expert in certain areas. So it's not worth it for us to bring people on. But other than that, not with the systems piece. And honestly, from the AI perspective, from what I've seen, these systems are not that expensive. Not really. Like, you can get a decent system installed for, I'd say 50,000 a year and then probably less going forward because you've hit the implementation fees and that could do the work of some ftes.
Elisabeth Schwartz [00:06:40]:
So there's some efficiencies to be gained with that.
Steve Swan [00:06:43]:
And so how many ftes do you have currently on your team?
Elisabeth Schwartz [00:06:47]:
Two, including me.
Elisabeth Schwartz [00:06:49]:
Yeah.
Steve Swan [00:06:51]:
You got a lot done.
Elisabeth Schwartz [00:06:52]:
Yeah, we manage about 40 vendors. I'm probably, as we discussed, looking to bring in somebody else to manage some of the pieces that I outsource. I think it would be effective to do it that way, but I haven't started down that path yet, so I don't know what the market looks like.
Steve Swan [00:07:11]:
Right.
Elisabeth Schwartz [00:07:11]:
Yeah.
Steve Swan [00:07:13]:
Always changing. Right. It just depends on what the particulars are, where you're going, what you're doing, and what you need.
Elisabeth Schwartz [00:07:20]:
Right, exactly. And then along with the AI piece, where I'm seeing life science companies use it a lot, is in clinical trial modeling, clinical trials, using the AI, what might occur given the inputs.
Steve Swan [00:07:36]:
It's so much data to get through and to churn through. Like you said, that's where the compute power can really save time, effort and money. On the clinical side, I've spoken to somebody who did let me think about this for a second. Somebody did some sort of marketing thing with it where they spun up some sort of platform that built a model that would send out some communications, and if they got back x, it would send out y and so on and so forth. Somebody else built up something with a large language model for their scientists to dive into research. Show me everybody's experiments having to do with x, y and z. I don't know the particular terms, but yeah, again, like my last guest said on my last podcast, we're kind of in the wild west. We're kind of rubbing sticks together and figuring this out.
Steve Swan [00:08:26]:
But we've all got to do it right. We've all got to crawl before we walk. That's just the way it is. I think some of the traditional systems that we're using, from what I'm hearing, are going to be challenged by some of this.
Elisabeth Schwartz [00:08:41]:
Yes. And where I think a big sticky point is going to be for other CIOs is security, because, for instance, I heard that they can replicate people's voices now. So you might get a call from someone acting as the president who sounds exactly like the president, saying, wire me $50,000. We've looked into some of these pieces and come up with some modifications to how we do business because of it.
Steve Swan [00:09:09]:
Well, yeah, the security and the data. Right. I mean, another big concern is all the data. I had another guest from me that was with me, and he started at a small company, right? And his company grew and it ballooned. And he said to me, I asked him, like I asked you, what's your one piece of advice if you had to think? He said, get your data ready. It's the gasoline for the engine. Get your data ready because you can buy outside data, but then you got to figure out whether you can trust that you have all this internal data. You can trust it.
Steve Swan [00:09:46]:
You just got to get it ready. Right? Are you seeing the same sort of thing?
Elisabeth Schwartz [00:09:49]:
Oh, absolutely. And that's the lift with these AI systems. That's exactly what it is. You've got to get the data in there and usable and have it be trustworthy. So, for instance, again, going back to the know, what's the meal limit for Chicago for dinner? You've got to trust that the engine is scraping and regurgitating information correctly in a way that is correct. It could be wrong if it doesn't interpret it correctly. And that's it. It's the data.
Elisabeth Schwartz [00:10:17]:
It's understanding the data. It's making sure that the model understands a question and then responds accordingly. And that's the testing you have to do.
Steve Swan [00:10:30]:
And I don't know the answer to this. And if you don't know, I'm just throwing this out. Do we know when it's coming up with a synthetic answer, when it's taking the average of some of this data? So sometimes what I learned about this just today, somebody was talking to me about how it takes data, and then it can come up with an average say, well, it's not this, it's not this. So let's extrapolate and guess it. It's this. And AI will do that. And I didn't know that until. Don't.
Steve Swan [00:10:57]:
Do we know when it comes up with the synthetic answer? I don't know the answer to that. So you don't know?
Elisabeth Schwartz [00:11:02]:
I don't know mean, but you probably heard the same thing I did where there was a lawyer that put a case studies, like other cases, for a complaint, and AI made them up. So basically the lawyer went in with these made up cases. Right. So to your point, I don't know. And that's a good question and I don't have a good sense of it.
Steve Swan [00:11:22]:
Right? Yeah, I have no idea. So now, when you have your data and your data sets, right, is there somebody inside that handles that? No, you use vendors. Right. So vendors are going to be handling that for you.
Elisabeth Schwartz [00:11:36]:
Right.
Steve Swan [00:11:36]:
Your data and getting that in shape for you?
Elisabeth Schwartz [00:11:38]:
I don't know the answer. The bottom line is we see this as a business lift, not an it list. So the business user is going to be responsible for working with a vendor, getting the information entered and qaing it, and we'll help. And it's exciting for us. We're happy to help and be involved, but we'll never be the expert the way the business user would be. And they are the ones driving these projects.
Steve Swan [00:12:02]:
Got it. Okay. They own the data. They have the data. The data is theirs. You're building the technology side of it, or. Yeah, your group is putting that again.
Elisabeth Schwartz [00:12:10]:
It's a light lift from our perspective. All we need to do is tie it into our office system.
Elisabeth Schwartz [00:12:14]:
That's it.
Steve Swan [00:12:15]:
Really?
Elisabeth Schwartz [00:12:15]:
Yeah. The AI isn't generating answers for how do you use office?
Elisabeth Schwartz [00:12:22]:
It's not.
Elisabeth Schwartz [00:12:23]:
It's answering questions that the business wants to know. And so they are the experts. They are the ones qaing the data, handling the data, and they're happy with that role. They understand that they'll get better result from that.
Steve Swan [00:12:37]:
Know, you bring up a point, know some people are saying, well, you're reading the headlines anyway. It's going to replace everything and everybody. And I think Microsoft named theirs copilot for a reason. Right. Because humans have to be involved with the whole thing. I mean, it's not going to take over everything for you. Auto drive is a perfect example. Right? The self driving car.
Steve Swan [00:13:01]:
It can't do it. I can't even figure that out. We're not going to be displaced, at least in the form that it's in today. Doesn't seem to me.
Elisabeth Schwartz [00:13:12]:
That's just it. We don't know.
Steve Swan [00:13:14]:
We don't know.
Elisabeth Schwartz [00:13:15]:
No. I could see it replacing some roles, but certainly not all. You still need humans involved.
Steve Swan [00:13:22]:
Absolutely. So now from a technology perspective in biotech, what are some of the trends beyond that? Beyond AI that you're seeing or some of the things that you think are coming up? That should be something folks are thinking about or seeing? Are we talking about crms, erps, accounting systems? Is there anything along those corporate functions or is it really about the AI right now that's really around the peripheral that's making a big difference for us or will make a big difference?
Elisabeth Schwartz [00:13:53]:
There's that and there's security. And the government's getting much more involved in how security operates. And if you're a public company and if you have a breach, you have to report it to the SEC and the FBI, I believe I'd have to look back on the reporting requirements. So a lot of what I'm learning now are just new government regulations and making sure that our policies incorporate them. I'm not saying anything new on the ERP side yet. I haven't seen a whole lot of system changes except for around security. Security continues to be a very talked about topic.
Steve Swan [00:14:29]:
When you go out and look for the person that you and I were talking about, what kind of skills, background, what kind of things are needed in an organization like yours for a role like that?
Elisabeth Schwartz [00:14:40]:
Being able to learn and being nice or being able to work with people. I think given how specialized systems are, I think as a manager, we have to expect that we'll hire people that will not necessarily have the experience we want them to have for the role. So we'll have to choose people that have the ability to learn and some proven track record learning and working on systems and then send them out to training to specifically handle things.
Steve Swan [00:15:09]:
Yeah, because you can train, right? Yeah, you can train.
Elisabeth Schwartz [00:15:13]:
Yeah.
Elisabeth Schwartz [00:15:14]:
If someone's motivated and reasonably intelligent, usually you've got a pretty good employee. It's kind of surprising how many people don't have those two pieces. But isn't it? Yeah, it is, actually. I know you see that. I know. Right?
Steve Swan [00:15:33]:
Yeah. I don't think my goals were too lofty when I raised my kids. I want to create productive citizens that can reason their way through their day without negatively impacting others. Finding folks that can do all three of those is a lot harder than it sounds. Right?
Elisabeth Schwartz [00:15:51]:
I know it is.
Elisabeth Schwartz [00:15:54]:
Yeah.
Elisabeth Schwartz [00:15:54]:
And I understand that I had some of that with my kids, that I wanted them to have the tools to make good decisions.
Steve Swan [00:16:02]:
Don't take down the house. Right. While you're doing it. Right.
Elisabeth Schwartz [00:16:04]:
Right.
Elisabeth Schwartz [00:16:05]:
Yeah.
Elisabeth Schwartz [00:16:06]:
Train them how to think, not what to think, necessarily.
Steve Swan [00:16:09]:
Well, also, be accountable. Fix your own stuff. Don't say, hey, I don't know, and move on. It's theirs. They created it. Or if it's not theirs, maybe they don't hand somebody a problem without. No manager wants to be surprised. And also, no manager wants you to dump something on without coming up without at least having a suggestion for a solution.
Elisabeth Schwartz [00:16:31]:
Correct.
Steve Swan [00:16:32]:
I see it all the time when you go out. I'm just trying to help you here as you're out thinking about your person. Right. Do you want somebody from a small place?
Elisabeth Schwartz [00:16:45]:
I don't care. Okay.
Elisabeth Schwartz [00:16:47]:
I really don't care. I want someone with the skills to learn, if they learn some of the subject matter that they'll need for the job.
Elisabeth Schwartz [00:16:56]:
Terrific.
Steve Swan [00:16:59]:
And now when you think about. We touched on this a minute ago, when you think about data, right. We talked about data and we talked about data feeding AI and things like that. Do you find yourself or does your business? I guess I should say that because. Right. They're the ones that own all this stuff. Do you get from them or do you hear from them that they're in need of different sources of data? Do they go out and find vendors and things like that for their data? Because I've heard from smaller companies, here's where I'm going with this. I've heard from other small companies that when they go out to buy data, they're forced on the imss and such to buy these huge sets, and they don't have the money for that.
Steve Swan [00:17:34]:
That's just not a thing. So they're out looking for vendors that have more niche sets of data at a lower price point. Right. Is that something that your organization has come into?
Elisabeth Schwartz [00:17:44]:
We haven't had that particular issue. You spoke about that yesterday, and you sound like you have a great source of somebody that can provide data for smaller businesses without the crazy expense that usually comes along with these data sets. But no, I think to be competitive in pharma, there are data sets you just have to buy, and they are expensive. You're going to want the doctors that prescribe in your particular field. You're going to want to have that database that's constantly refreshed to see who's active and et cetera. Those are expensive data sets, but they're required mostly for the pharma industry.
Steve Swan [00:18:19]:
You have to do it. I'm shocked that there's not a, I don't know, in our business, some sort of, I guess you can't recycle, you can't share those resources because it's the data that you got to take in the data and use it how you need to use it. And you can't let other folks know how you're using it because you're doing your thing. Right. And there's also FDA involved with that. So there's no shared sort of resource there. You got to do it on your own. I know I'm thinking crazy, right?
Elisabeth Schwartz [00:18:50]:
Very expensive to collect, collate. In fact, I've read that, like, Iqvia does not make a lot of money off of their data. That's not where they make the money. Yeah, I've read that again, I try to source it for you. I don't want anyone to run back and tell this to somebody, but I've read this, that that's not a big money maker.
Steve Swan [00:19:14]:
That would explain why several years ago they were pivoting or trying to pivot right more to the system side.
Elisabeth Schwartz [00:19:21]:
Yeah, I think that's why I read that and I'll see if I can find that article back on that.
Steve Swan [00:19:27]:
I remember reading about them pivoting to the system side, and I'm like, why? They're the 800 pound gorilla data. But I guess thinking about it, why would you pivot from anything? Because you're not making it. You're not pivoting from something because you're making too much money. Right?
Elisabeth Schwartz [00:19:39]:
I think that's right. And it explains some of know there's battles going on between Viva and Iqvia with data. There's lawsuits now, and they're all very sensitive about who gets the data, how much they're paying for it. What are you doing with it?
Steve Swan [00:19:53]:
So what else would you like to share or talk about? I don't want to take anything from you where you want to chat about anything in technology or business or pharma, as we continue to talk here, anything that you're.
Elisabeth Schwartz [00:20:11]:
No, I think you've hit them again. Being in a small to medium sized company. The way we manage our department is different than someone at a large company would. I know where the in house people, it's very much of a vendor model, but I would expect as we grow, we'll bring more people in for the areas where we use a lot of vendor talent, because it gets expensive to constantly outsource to vendors.
Steve Swan [00:20:36]:
Yeah. I've seen companies, even up to several hundred million dollars in sales, they'll have one person that's responsible for, I don't know, one security, one infrastructure, one commercial, one corporate or ERP. You know what I mean? Kind of doing that and then somebody that. Each one of those individuals may have strong skill sets, even architect kind of skills, but great business skills as well. Kind of the high level business relationship manager, ITBRM. Almost. Right. And be able to go between business and their vendor manager, their vendors.
Steve Swan [00:21:10]:
And once they get big enough, then once they get beyond that, then they've got to deal with a lot more. But they've been able to sustain that model for a while as they continue to grow.
Elisabeth Schwartz [00:21:22]:
Right?
Elisabeth Schwartz [00:21:22]:
Yes.
Elisabeth Schwartz [00:21:23]:
Well, that's probably where we'll be if we continue to grow.
Steve Swan [00:21:27]:
All right, well, I always like to ask one final question before we part ways, and it's more of an off the cuff question. I didn't mention this to you before. What would you say would be your favorite live band you've ever seen in concert? There go the eyes.
Elisabeth Schwartz [00:21:47]:
Okay. I'm a nerd. I don't go to concerts. Too many people like too loud.
Steve Swan [00:21:53]:
I get that a lot, by the way. A lot of people say the same thing.
Elisabeth Schwartz [00:21:57]:
Yeah. I mean, I'm older, really?
Steve Swan [00:22:02]:
Even back when you were 15? I don't know that I liked.
Elisabeth Schwartz [00:22:07]:
Probably like Billy Joel. I mean, can't go wrong with Billy Joel. I've seen him a couple of times.
Elisabeth Schwartz [00:22:12]:
Yeah.
Steve Swan [00:22:12]:
No, he's awesome.
Elisabeth Schwartz [00:22:13]:
He's a good performer.
Steve Swan [00:22:15]:
He's a great performer. We didn't see him till last year, so that was the first time seeing him.
Elisabeth Schwartz [00:22:19]:
Oh, wow.
Elisabeth Schwartz [00:22:20]:
In the city.
Steve Swan [00:22:22]:
We saw him in the city. Yeah. So we had tickets to go the year before, two years ago, on my wife's birthday, they canceled. That was in December, and he didn't play till June of last year, so it took six months to come back and do it. But it was great. We took our daughters, who are 23 and 21, and the four of us went. It was fun. It was good.
Steve Swan [00:22:41]:
My wife loved.
Elisabeth Schwartz [00:22:42]:
I'm glad.
Steve Swan [00:22:43]:
Yeah. I think he's wrapping it up, though, at the garden. If I'm not mistaken. I think he's kind of run out of Runway there.
Elisabeth Schwartz [00:22:53]:
Energy. Yeah, a lot of energy.
Steve Swan [00:22:56]:
He's a little older than us, right? Yeah, a lot. Anyway. All right, well, thank you very much for your time. If anybody needs to get in touch with you, one of us, you'll see our links there on the podcast. And thanks for joining us.
Elisabeth Schwartz [00:23:16]:
Thanks for the opportunity.