What If Your Data Could Speak? The AI Revolution in BI. An interview with Nico Frantzen.

Nico (00:05):

The challenge is how do you create a solution that scale across the organization that can understand it. Your business, your data models that can speak the language of your users. They can understand your product, they can understand your customers. Doing this will require, you some level of effort to bring all the right experts in the room.

Jim (00:24):

Welcome to What If? So What?, the podcast where we explore what's possible with digital and discover how to make it real in your business. I'm your host, Jim Hertzfeld, and we get s**t done by asking digital leaders the right questions. What if, so what, and most importantly, now what?

Jim (00:41):

Hey everyone. It's Jim. I'm really excited to be here with Nico Frantzen. Nico, welcome to the show.

Nico (00:46):

Thanks for having me.

Jim (00:47):

Nico, tell us a little bit about what you do here at Perficient.

Nico (00:50):

So, at Perficient, I'm a director in our data and intelligence group. I lead our AI ML practice for the past six years.

Jim (00:58):

Great. How'd you get into that space? Just luck?

Nico (01:02):

Not luck. I was primarily doing content management and workflow applications, and I got into the space of how to figure out to extract data from documents. And I got really interested into natural language processing and naturally evolved toward that space.

Jim (01:19):

And taking some really kind of old school, tedious, been around forever kind of things and automating. Some of those activities, right? That's what it's about. Well, for the last year, it seems like the only thing that people want to talk about is generative AI just sort of leaped into the collective consciousness, not, not just in the industry, but,

Jim (01:40):

You know, everywhere. I mean, I think as we record this, we were coming out of a Hollywood writer's strike, and it was an interesting strike because they felt very threatened by it. So it's really a kind of touched on all aspects of our lives and our culture.

Jim (01:52):

So, but, you know, a year into it, there's a lot we've learned, invoked a lot of ideas and imagination, but you know, today, how do you define generative AI.

Nico (02:01):

So generative AI is part of, you know, the broader AI space, but it's really, you know, AI capable of generating content whether it's text images, other media leveraging what we call, you know, generative, you know, models really, those models are just like any other traditional machine learning models, but they're really there to learn from your pattern and structured, from their input, like training data.

Nico (02:25):

And trying to replicate and generate new data with similar, you know, characteristics, it's just traditional machine learning on steroids; it's been, shown to be very powerful.

Jim (02:34):

Yeah, I guess the term generative, like on its own, we know how to generate things. It's not just generating images or sound. There's a lot of things that can be generated, and I think the interesting thing to me is that.

Jim (02:46):

When we say generated, we're talking about creating things that just didn't exist, which is pretty cool. And I think that's why it's so exciting. But you wanted to talk about something we're calling generative BI, generative business intelligence. So, how can you take this technology and generate intelligence or look into that data? But I don't want to mess that up.

Jim (03:08):

So maybe, in your words, you tell us what is generative BI.

Nico (03:11):

Yeah, we, I don't think we have coined the term, you know, quite yet.

Nico (03:14):

But generative BI is really generative AI applied to business intelligence. Think of your traditional BI applications dashboards.

Nico (03:21):

How can we leverage generative AI, you know, in that concept? The idea is to provide a more natural machine-to-human interface. To access and surface your insights from that structured data, my personal opinion is that it's going to completely change how we think about traditional BI.

Nico (03:38):

You know, historically, you think of BI. As you know, querying data, you know, dashboarding, it might take teams of data engineers and BI Engineers and tools, you know, to put that together. But with generated BI - I think we're really gonna lean into a new era of storytelling. How can you formulate A question a hypothesis, get the relevant results, display them in a way that's meaningful to understand, and then generate key insights that you can explain, right, and easily share and collaborate with other people.

Nico (04:10):

Asking questions is natural to us. Looking at a dashboard is not natural. You still have to understand the context of that data. What was the intent originally? So when you see all of this put together, you ask the question, you see the data, and then you can explain what you're seeing all in one go. I can share that with a peer of mine, and we can easily collaborate and understand and get the same insights from that data.

Jim (04:36):

And that's been around for a long time. I can remember a time where there was something called green bar. It was a printout, like a dot matrix printout. It was green and white, so you kind of find your way through it.

Jim (04:47):

I remember hearing stories about, every night, we run a job in the mAInframe, and it prints out, hundreds of pages. And then, I go through it to find the one thing I was looking for. You know, it's just the way it was done, but We always want to be data-driven. Sometimes, we really don't want to see the data. Nobody wants to see bad news, but you want to be able to action on it. You want to make sense of it. We're generating data with systems all the time. It's been, you know, a hallmark of computing, you know, and IT departments from the beginning.

Jim (05:16):

But why is this approach important to companies today? What's in it for their customers? What's in it for their internal users and the business leaders in the companies you deal with.

Nico (05:27):

I think maybe just to go back a little bit between the differences between your Gen AI and, and, Generative BI is. A big scare around generative AI is the fact that it can create answers, Create a new truth or a new reality that's just not real, people are very scared about that. On the other end, you have data in your enterprise that tells exactly. You know, what happened? What are your cells? What are your Customers saying? And as a business, I want to understand and get answers to my questions that are accurate and truthful. And so generative BI can actually leverage gen AI, but actually ensure that we're representing and telling the truth about, your organization.

Nico (06:10):

And so this technology is really going to revolutionize how we access data across the organization. We're going to democratize that access. We're going to enable anyone in the organization to be able to make faster and better decision making,

Nico (06:25):

In a self-service capability, which is the real benefit here.

Jim (06:29):

So, yeah, it's interesting. You brought up the idea of is it real, right? And so we're seeing very tangible, consumer oriented results of generative AI. And we're all questioning is that real? And again, that's not a new problem, right? Photoshop's been around for awhile.

Jim (06:41):

Is that a real photo? It was, I think it might've been Photoshopped. But yeah, the idea of the truth and what it means. Going back to something Mark Twain said, there's lies, damn lies and statistics. It's an interesting problem, But solvable in your mind, right?

Nico (06:56):

Solvable and with real business value. There's always been that bottleneck you just mentioned it in your example. It takes sometimes a village to put together the right dashboards for executive leadership, for management it takes teams of BI engineers and data engineers that often get bombarded with questions and they have to scramble to source the data, you know, work together to shape it and then display on dashboards, That takes sometimes weeks of gathering the requirements, developing it, testing it.

Nico (07:26):

And then finally, you share with your executive and guess what? They're like, well. It's not exactly what I was looking for. And you repeat the whole cycle all over again, right? Because the typical waterfall approach to asking a question and doing all this work to generate the answer is very tedious, time-consuming,

Nico (07:44):

Costly to the organization. And often by the time you get the right answer, that question is no longer relevant. You moved on to the next problem right in your organization. You have different questions. And so you need really this idea of generate BI where you can self serve and ask questions and get, those answers, quickly without the need to involve, you know, teams of engineers, you want to put that together is really, you know, where the business value is going to land.

Jim (08:08):

Well, and you know maybe I missed it, but you didn't bring up data science, right? That's kind of, I'd say relatively new concept. And I always thought of data science. It's a great, phrase, by the way. It's a great, it's a great label, You know, because you have.

Jim (08:21):

Company. I always thought of as, you know, you have, data people who understand the insights, understand the business. Maybe you can understand the voice of the executive and what they're looking for and sort of the science and engineering behind it to make it happen.

Jim (08:34):

But the nature of analysis to me is, iterative. It's exploratory, So you do have to do a lot of what if. And I think if I'm hearing you right, the bottom line here is we're moving faster, we can get to that analysis faster, and then we can get through that analysis faster.

Jim (08:51):

Is that maybe a good way to summarize?

Nico (08:52):

That's right. The world is changing faster. I mean, think of when COVID happened, and every company was scrambling to figure out how they were going to react to it. That type of technology would have allowed them to get the answers, to get the insights of how their business was running, was functioning make faster decision to react more quickly to how their customers needed to be served.

Nico (09:14):

I think with the science in generative BI, I think there's also this future era of exploration where you can do some what-if analysis. Hey, if my market, you know, it goes down and so you can leverage now traditional, data science inside of combining it with your actual data, to get some predictions about what might happen in different scenarios, again, without the need to involve teams of data scientists data engineers or BI Engineers to do that.

Jim (09:42):

Kind of a self-service element to it. We know this is all evolving quickly. You've been involved in some pretty significant, pretty compelling uh, applications and, projects. Maybe share one of those with us, maybe kind of a real-life example of how this velocity is being created.

Nico (09:58):

Yeah. So first of all, I mean, it's not a new concept. Many vendors have started to really think about the potential value that it can bring to the table we've seen products like AWS, Quicksite, Databricks is coming out with their Lakehouse, IQ, you know, features where, again, it kind of learns about the data, learns about how you model your data can quickly help any user without any sort of technical skill, to access that data and, represent it, quickly.

Nico (10:27):

I remember building, a few years back, way before Generative AI was a thing, leveraging existing technology like natural language processing, a really compelling prototype to show customers, you know, the power of this concept. Nowadays, you know, with Gen AI in the mix I'm working with actually one of the top pharma companies to see how we can leverage generative BI to explore clinical data and to improve patient outcome through better analysis of the efficacy or the safety of, particular, clinical protocols.

Nico (10:57):

And so think of, again, COVID, right? You had probably the largest clinical trials going on that needed to happen like super fast. Everybody was, rushing out the door to get answers and understanding on those large populations how blood thinners could affect the use of the COVID vaccine.

Nico (11:18):

At the time, it required for them to stop and pretty much put on pause every other clinical trials just to put more bodies to analyzing, you know, this huge amount of data. It was just brute force, And that's how they got it out the door. And probably know that COVID is going to happen again in some form of another, right?

Nico (11:37):

I mean, those situations are going to happen again, and pharmaceutical companies are going to have to react much faster because now the expectation is, is higher; we need a vaccine in like six months or less.

Jim (11:48):

Yeah.

Nico (11:49):

Boy, it was three years, four years, right? We had the time, you know, to do this analysis, but now we have to do it much faster.

Nico (11:55):

And so these are the kinds of real problems that we're trying to solve and, putting in the hands of physicians within pharmaceutical companies to answer some critical. Medical questions about the patients that are doing those clinical trials.

Jim (12:09):

By the way, that's, that's an amazing application that, you know, literally affects hundreds of millions of people. But if you're not at that level, or maybe you don't have the resources, what can other companies do to get started and understand or apply Generative BI.

Nico (12:26):

So it's definitely an evolving space. Many vendors are thinking about it. I don't think we're quite yet at the maturity phase. I think there's some fairly challenging problems, especially at scale across the organization to solve. But I would advise, you know, there are some compelling.

Nico (12:42):

It's kind of off-the-shelf, you know, like a product that you can start leveraging, and it's all about demonstrating value within the organization, iteratively. You can't boil the ocean. So, you know, start small and, you know, use those of-the-shelf products to show how you can enable data-driven decision-making.

Nico (13:01):

I would also advise companies to really start thinking about this problem more holistically, right? Bring together their business stakeholders, data engineering, their BI teams, you know, together along with if they have any right, Gen AI center of excellence, companies are spending a lot of time trying to figure out what they can do with Gen AI and in my opinion, this is a very compelling use case. The value is real. The technology is readily available to solve that problem.

Nico (13:29):

The challenge is how do you create a solution that scale across the organization that can understand it. Your business, your data models that can speak the language of your users; they can understand your product, and they can understand your customers. doing this will require, you some level of effort to bring all the right experts in the room.

Jim (13:47):

Yeah. I feel like a lot of new technologies or, you know, there's a hype cycle, and, you know, where it peaks, I think it's called, you know, the peak of inflated expectations and the. Trough of disillusionment. If you guys have been in the industry, I think that's kind of the Gartner model.

Jim (14:01):

But, you know, I'm waiting for that to happen. But you haven't really gone that far in the organization with, generative AI, BI is probably a great place to start because it sounds like you're going to get some really tangible business value.

Jim (14:12):

Pretty quickly. And there's a client, I think we're working with Nico that doesn't need a pretty hard ROI before they, you know, ready to invest in anything. So I think that's great advice to get started. Nico, thanks for joining. Love talking to you and I love what you're doing. Keep it up.

Jim (14:27):

I know you're learning things every single day, and I can't wait to see what you do next.

Nico (14:32):

I’ll let you know in a few months when we have a prototype ready to share.

Jim (14:38):

No, not a few months in the public.

Nico (14:42):

In the public.

Jim (14:43):

All right. Thanks, Nico. Have a great one.

Nico (14:46):

Thanks, Jim.

Joe (14:46):

You've been listening to What If? So What?, a digital strategy podcast from Perficient with Jim Hertzfeld. We want to thank our Perficient colleagues J. D. Norman and Rick Bauer for our music. Subscribe to the podcast and don't miss a single episode. You can find this season along with show notes at perficient.com. Thanks for listening.