#264 From Gen AI to Gen BI with Omri Kohl, CEO and Co-Founder of Pyramid Analytics
48:28 Share

There's been a gradual progress in BI. I think generative BI is going to be a quantum leap. I actually believe that it's going to allow everybody to use insights to drive their business. If it was difficult and cumbersome and annoying and unexplainable, how you got there, what to do with it, I don't want to use those BI clunky tools. All of a sudden, with those generative BI systems, you can actually give it to everybody.

Welcome to Data Framed. This is Richie.

As with every other piece of software, generative AI features are rapidly being added to BI platforms. If you're a regular listener, I'm sure that's not surprising news. But since generative AI combined with BI, or generative BI for short, has some big implications for a lot of data roles and also for consumers of data, it's worth thinking about the details.

If your job involves creating dashboards, you might be asking yourself whether these features will make you more productive or let you build better dashboards. You might even wonder if you need to bother creating a dashboard at all if your users can just give a dataset to a chatbot and ask it questions. On the other side, if you look at dashboards at work, you might want to know if AI is going to help you get insights faster or make better decisions.

And on either side, it's clear that if you work with data, your job is going to change in the near future. So you probably want a sense of how it's going to change. To answer your questions, naturally, I have a great guest for you. Omri Cole is the CEO and co-founder of Pyramid Analytics, a decision intelligence platform company.

He spent 16 years growing the company and frankly, since Pyramid's main competitors in the BI space are Microsoft, Salesforce and Google, keeping the company competitive is no mean feat. At this point, Omri is very much a thought leader in BI and if you want to hear more of his insights, after you've listened to this of course, you can check out his Data Shark podcast and the associated Data Shark blog. So with that, let's find out all about Generative BI. Intro

Hi, Omri. Welcome to the show. Hey, Richie. Thanks for having me. I'm pretty excited to share the next hour with you. Absolutely. I'm looking forward to it. But to begin with...

I was thinking that there are actually a lot of really great BI solutions out there. A lot of the platforms just kind of work. So what's left to innovate in the BI space? So first of all, I'm not a future teller, so it's hard for me to tell you what's going to be. But I think when you look at the evolution of data and analytics is a big part of that, I

I think that we're definitely at the maybe beginning of or the middle of the beginning of the next generation of analytics tools, which refers to AI. So I think, A, lots of things to innovate in around access to insights and creating insights. That's one side. I think the other side is getting to the data asset. Data became the next whatever goalpost

A golden era of data is today, and people refer to it as the next maybe oil. So drilling into the data or harvesting data or collecting data is definitely another area where lots of innovation will have to happen in order for those two big tectonic changes to happen. On the one hand, getting people to build insights from what they collect in their companies. And the second part is how do you actually

manufacture the data to make it usable. Okay, yeah, that's really interesting. And I like the idea that the next sort of evolution is just about making data more accessible so everyone in your company can get those insights.

And it's also interesting that it's a kind of multi-stage process. You need the data processing up front, and then you have this BI layer at the end, and then that's how you're going to get the insights. So it's like, oh, it's a whole flow. So you also mentioned AI. We're definitely going to spend plenty of time talking about AI, particularly generative AI. How can you not? I know, it's just invading everything. But yeah, are there any other trends you think that aren't related to AI that you think are particularly notable?

I think that one thing that is not even related to technology when you talk about data-driven culture, it's actually internal behavior. How can you transform people that are very much driven by intuition, they're driven by their experience, they're driven by something

sometimes conversations with their peers and colleagues to ignore that noise and look at the data and drive their business from that lens exclusively. And this is one extreme.

Obviously, the other extreme is completely ignore data. So how do you augment the two behaviors together? On the one hand, build top-down and bottoms-up data culture. So management will only discuss, I don't know, sales performance.

once they have the data. We're not going to talk about your aspiration. We're not going to talk about your expectations. We actually want to look at the data and then we can discuss what's the additional layer on top of that. So that's top down. And bottoms up, how can I in my organization

make sure that the next decision I'm going to make, and people make sometimes, you know, hundreds of decisions a day in their line of business. How can I make the next decision using data? Is it accessible? Is it accurate? Can I trust it? Is it available for me? What do I need to do to get access to that data? So I think that embracing that culture first

Without that, there's no point in implementing any analytics tool. Yeah, so process is always like the hard part and working with people, like generally much harder than working with technology, I think, a lot of the time. You know, data literature is probably the most important, difficult first step.

to even starting to appreciate the importance of becoming a data-driven organization. Absolutely. And so you mentioned the idea that sometimes you have people who are completely driven by intuition. They don't look at the data. The other extreme is like, well, we're not even going to talk about what we think until we've seen the data. Is there like a happy medium and how might you go about implementing it?

Yeah, I mean, first of all, I think that happy medium is proportional. My happy medium might be very different to your happy medium and to someone else's. I think we need to have a very critical relationship

You need to start using data and start making decisions and then look back and say, okay, that decision about that data for that line of business was successful to that degree. How can I improve it? How can I fix it? So I think constant feedback and constant feedback into the data and into my decision making is the first step to building a much more successful business.

analytics implementation. We're talking about BI and we're talking about AI and those are amazing tools. But eventually, if you don't use those tools properly and you don't utilize the insights that you will receive from those tools, they're useless. It's just a tool. We eventually will control the outcome and the narrative that we will be creating using those systems. I like that. And just thinking about my own experience,

When things are going wrong in some area of the business, that's when I tend to notice my intuition is probably wrong. And that's when I tend to start pushing for like, well, I need to bring more data into this. So do you have a sense of like what the point is where you go, okay, we need more data here? I'll take it to the macro level. When companies are...

investing the most in their data assets in times of crisis. So, you know, macroeconomic crisis is maybe the first step for people to say, "Okay, I need to understand what's going on in my business. Where am I bleeding? Where am I successful? I want to double down on the areas that I do well. I want to shut down, you know, line of business that don't perform." And kind of when you're cruising along, you couldn't care less. It's fine. Everybody's happy.

But in a disastrous moment, in a critical moment, in downturns, that's when you actually want to have super visibility to your business. And I think that it's really interesting to hear you say, kind of reflecting about your own experience, but think about it from a global perspective. When you look at companies holding off on technology spend,

It's true, they will stop spending on all kinds of tech stuff, but they will increase their spend on data and analytics. Why? Because eventually there is a very clear ROI to implementing and using properly data and data analytics. So I think that those moments are definitely defining moments for our space.

Absolutely. So yeah, when you've got some big disaster happening, that's when the decisions really are critical. You need to make sure you've got the right answers. That's where the data-driven side of things comes in most useful.

But I guess that leads to a problem. So if you are waiting until there is a disaster and then you're suddenly like, well, I need the data, it feels like maybe that's too late. Do you think businesses need to be investing in data before they get to that crisis point? Well, first of all, I'm not objective. I want everybody to buy from us, right? So I would say buy anytime, all the time. But I think if you're only starting in the moment of crisis, you're too late to respond. You want to make sure that

At least you built the foundation to be able to act on time in good times and bad times. It's not necessarily that you need to prepare for the catastrophe. You should also be prepared for good times where you actually see one line of business exploding and you actually want to go and double down on it. How do you do that? Where? Why? Why does it happen? Which product? Which territory? Is it the right salespeople? Is it not? Is it my marketing department?

so many questions. Is it my engineering group that created some ingenious idea and everybody just wants to jump on that product? So I think that it's a constant investment

If you look at the most successful companies in the world, they're eventually data-driven companies. If you look at Google, you look at Amazon, you look at Facebook or Meta, even Tesla in manufacturing of cars, what made them so successful is the ability to collect data, to do something with the data, to build those autonomous drivers. So all of that eventually is data.

And if you want to be ahead of your pack, invest in data. That's what you should take from all that hand-waving story. Data is the driver for every successful business today. Definitely with you. If you're going to be successful, you do need to get your data game. Yeah, absolutely. Wonderful. So we're going to have to talk about generative AI. But I know, Pramod, you have this term generative BI. So can you just first of all explain to me what is generative BI?

Absolutely. So, you know, Gen AI is something that everybody uses. My kids use it. My friends are using it. My colleagues are using it. People create homework and people build, you know, new songs and do lots of stuff.

And we were thinking, how can you take that experience but to the business world? Why is it a problem? It's a problem because my company's data is not public. It's actually my asset. I'm not ready to open my banking system to one of the LLM creators if it's Mistral or OpenAI. I don't want to insult anybody else. So all of them, I don't want to give them access to my assets, but I still want to leverage those LLMs.

So we were thinking, can I actually take the public LLM from one of those big companies, take my private data from my company, connect the two together and allow me to have a journey that is secured, governed, enterprise ready, that is not exposing me to any risk outside of my gate, my environment.

And that was the idea behind all of that. And we started building out all kinds of solutions. And eventually we came up with what we call GenBI. And the idea behind it is that we're using LLMs for free language. You can go into your analytics infrastructure using Pyramid and basically start asking questions.

Tell me how my marketing department is performing for the region A, for product B, for campaign X. And you'll start seeing, and then I want to see it in a pie chart. No, you know what? I want to see the scatterplot. Can you add the linear regression as well? Absolutely. And that kind of flowing conversation, which were very common to do in the public space, is happening in your region.

private governed data. And I'll say one more thing that one of the challenges in analytics and specifically in BI tools

is that there is an adoption challenge, right? When you look at the adoption rate of BI tools across companies, it's, I want to say, in the 20%-ish mark. You should ask yourself, why not anybody using analytics? It's almost surreal that people still are making decisions from their gut feelings.

And because those tools sometimes could be intimidating, they're hard to use, it's not for everybody, not everybody knows what to do with it. And I think that the idea of GenBI is adding more and more personas, more and more of that audience that it's not so intuitive for them to start embracing GenBI.

analytics tool. It's easier for them to email the analytics person, hey, can you send me my report on my yada yada stories? And they'll get it. And GenBI is the attempt to make that journey for them much more accessible.

It does seem like there's a very natural sort of marriage between using generative AI and more traditional business intelligence. So it seems with business intelligence, you've got two distinct groups of users. You've got the end users of your dashboards, and you've also got the creators of the dashboards as well, and they can lead different things. So maybe we'll go into a bit more detail on how does generative AI help people who are creating dashboards?

So I think, again, it's the skill set, right? And we're all acquiring skills as we kind of go along our career and you learn more stuff. And, you know, sometimes you do it yourself. Sometimes you lean on other functions. And I think that generative BI is

allows the entry point to those tools to be much more accessible and easier. I don't think that if I'm like the top notch, you know, data scientist, and I love to write my Python, and I love to do those joints and connections and all that fancy stuff, which is incredible,

I'm not too sure I will tell my genie, hey, please build the next milkshake for me and I'll just drink it. I'll probably want to be in the kitchen and do it myself. But I think that it just lowers the bar

for more and more creators. And that's why it's generative BI. Eventually, we're trying to make creation of those dashboards, those narratives, those insights much more accessible for everybody. So in a way, we're almost trying to narrow the barrier between those two groups and tell those people that are

not necessarily thinking about themselves as the dashboard creators, that there is a way for them to actually embrace that analytics journey by themselves. And maybe for the more advanced users that are

are what you call creators or developers of analytics tools, this is just a supplementary or complementary solution for them. At least you want to start somewhere. So instead of all the kind of legwork that you need to do to build your first dashboard, imagine that you're heading to a board meeting and you need to build a PowerPoint for the

The board, if someone gave you the first draft and told you, all right, start working off that rather than from scratch, then, okay, it makes your life easier. So I think for the very basic people, it just allows you to do much more. And for the advanced people, it maybe takes away some of the mundane prep work that you would need to do.

Okay, yeah, that certainly seems to reconcile with my experience where if I know how to do something, I can just do it. But if it's something where I've not done in a long time or something that's completely new to me, that's where Journey to the Eye is really helpful because it can give me that at least the first step.

Okay, so switching to how it benefits users, you mentioned the idea of just being able to ask questions about your data set and it gives you an answer. So that sounds very much like a sort of chat interface rather than a dashboard. So do you think that's the future? Dashboard's going to go away now? I'll go even one step further. And I think the bigger question is,

Would AI kill BI? I think that's eventually maybe where we're headed. So first and foremost, I'm not so sure that the dashboard is dead, like the password should be dead. I actually think that dashboards are still part of the way we consume data.

rather than just simple numbers. We like to look at visuals. We like to see red and green. We like to see bad and good. We like to see a thumbs up or a thumb down. So I think that the graphic representation of our...

performance will stay. And I think it's actually a very useful way to have maybe a first snapshot of my business. I think that the drill down and kind of the further exploration and starting to ask questions and starting to go through that rabbit hole of your data and your analysis will be, I think, simpler and easier

and maybe for everybody, once you can do it with a chatbot or, you know, we have, and I encourage people to go and search on YouTube for Pyramid GenBI demo, you'll see us actually talking to Pyramid and asking voice to text to analytics questions. Some of them are complex questions, like do that math for me and do this and do that. And I think that might completely change and revolutionize the way

people actually interact with data.

Okay, so it sounds like there's a difference between exploratory data analysis and explanatory data analysis. So if you're just exploring your data set, you're just asking lots of different questions, you're not quite sure exactly what you want at first, then the chat interface makes more sense. But I guess for something like your core business metrics, you're always going to want to know what your monthly active users are or something. And that's not going to change. Maybe the dashboard is a better solution in that case. Is that sort of about right?

Yeah, I think you nailed it. I mean, I think that for my first glimpse into what's going on in my business, show me a dashboard, show me happy face, sad face, show me I'm fired or I'm getting a raise. That's all I need to know. But if I actually want to further understand why I got fired or why I got a raise, then I think a chatbot experience or a voice to text experience or a

Anything that is much more intuitive, almost the conversation that you and I are having, why not have it with your data? Okay, that certainly makes sense. And I just feel like there's always been this sort of gap with BI where you create a dashboard, but actually what you want to get to is maybe making a decision or finding out some reasons why and the dashboard has not been able to provide that. So can you just talk me through like,

How you close that sort of last mile? Like, how do you go from, here are some results from data to this is how I make a decision? We're definitely, we believe in decision intelligence, not just decision making, but actually we call it DI. So too many DI and AI and LDI. But the idea of decision intelligence is that eventually there is a constant feed or feedback into your data.

We made a decision, it did something, it feedbacks to us, tells us what improved, what didn't improve, and we continue to do that. And maybe connecting it to AI, I think that AI could definitely give you constant feedback. Not just when, I think when we're saying AI, people think about the chatbot, but AI can, or machine learning or non-rule-based automation can happen in multiple layers of your data stack, of your analytics stack.

And we can build semantic layers. We can acquire clean and fixed data with lots of machine learning and deep learning. We can create those semantic layers on top of that and build you a map of your conversation with your data, not verbally, but build your mapping of your schema. Then we can create automatic insights. So I think AI and non-rule-based automation exists across your entire data stack. And

Once you have that implemented, then decision making becomes almost an inherent part of that. Why? Because let's say that to me, sales is almost the easiest example. So, you know, I'm seeing one region of mine kind of dropped their quota and they're unable to sell. And I want to understand why. And one of the reasons is that maybe it's too expensive for that region. Why? Because maybe people in that region will spend less

on buying something. So if it was a good system, it would tell me, listen, it seems like you're losing on pricing. Why don't you adjust your price list? We call it in our space prescriptive analytics. But basically the idea is can I prescribe a solution to the insight we found? The insight was you're losing business. The prescription was change your pricing.

I did it. Fine. I listened to your fancy machine that told me to reduce the price. I did it. And guess what? Nothing happened. I continued to lose business. So either, you know, the prescription wasn't good. The data that I was using to make that decision was not great.

Or there are other parameters that are the reasons why I'm unsuccessful in this region. And I think that back to the conversation we had at the beginning of data culture and am I going to listen or not going to listen? I think that decision making using systems

is a discipline. You really need to decide that you're willing to go about taking your analytics journey to the next level and actually drive your business using those insights. And I think that AI could eventually, though it might sound like a weird comment, but could eventually add more comfort to people leaning on those tools to make decisions.

Why? Because it almost feels like I'm talking to you. I'm almost getting insights from someone who cares, someone who actually understands the business, understands me. And then when he tells me, okay, reduce the price,

And I do it even faster and smarter. And then maybe it will work. Yeah, so I really like the fact that you can make use of the subjunctive AI chatbot to sort of ask questions. They can give you advice. The tricky part is when the AI doesn't have as much context about your business as you do. So you mentioned the idea of, well, does reducing price work well or not? Unless the AI understands the state of the market and why your price is the way the price is,

it may just make stuff up and then because it speaks confidently whether it's right or wrong, you're not going to know the difference. So can you talk me through how you might go about like making sure that AI has the right context in order to give you a sensible decision?

Yeah, first of all, I don't think that there is an easy answer to that. I think that AI is in its infant days and there are lots of hallucinations and lots of reasons why we should use it carefully and we should decide how we're going about the usage of the insights and that stuff. So I think that

From a technology standpoint, lots of systems have been built to try and reduce the risk of using AI. RAG is one idea behind it, kind of framing the way you use those underlying data assets to make sure that you reduce the mistakes, if you will, or the hallucinations. So that's on the tech side. I think that given that it's still humans to machines,

We also need to be somewhat critical about what we're using and how we're using and what we're doing with the insights that were created. And should we actually fall in love with it and use it 100% and I'm only going after my chatbot, what it tells me, it's the words of heaven or nothing.

I'm augmenting my experience. And I think that today, augmented analytics is probably the most common practice that companies will embrace. Yes, we're giving you tools that will make your life faster, smarter, better. You'll be able to understand your business in a heartbeat rather than sometimes it took you, I don't know, months to get some insights on what's going on in your business.

So we made it super fast, super accessible, super valuable, but also it adds lots of responsibility on you.

to make sure that you're going to use it in a proper manner, that you're going to continue to question the insights that you receive. So I think my advice to everybody that is using AI or AI in our space is to augment your experience with what the automation is kind of driving towards.

Okay, so this is a fairly frequent topic on Data Framed. It's like, do you want the AI to augment your own capabilities or do you want it to replace you? And there seems to be a divided opinion on whether AI and humans should work together. Actually, that's maybe the most common opinion, but the other opinion is, well, long-term, we want complete automation. We don't want to have to have humans in the loop at all. I was wondering where you stand on this. Are there any specific cases where

It is better to have that human plus AI combination when you want to completely automate and go AI only or software only.

So I think that, you know, kind of the core story is what is the use case? It is use case based driven to take two very extreme ideas. One, would you put your life in the hands of an AI or a robot? And on the other one, would you let an AI make a decision on what's your meal for lunch today? And I think, you know, it may be

silly ideas, but eventually behind it is how much of the decision would you leave in your hands or in the machine hands? And I think that eventually what we're looking at in business is mission critical decision making and or mission critical application. So if it's going to fundamentally impact

the overall business. I think I would do what we typically do, A/B testing, further exploration, cross-pollinating of data and data assets coming from various places. I would do a thorough research before I let the AI decision to drive my next step.

And maybe eventually I learned that it said exactly what I was finding out from my research. And then with that confidence in what the AI is doing,

But I'll definitely not fully augment or not fully allow my AI or my data or my analytics to run the decision making. And if it's, you know, what should we order for lunch? Roll the dice for me. Like, you know what I like? You already know me because I shared with you my entire life on Facebook and Twitter and who knows what. And you know my preferences.

I'll roll the dice and I'll be very happy to be surprised with what you ordered for me. So I think it's really about the use case. I think that there are lots of practices and processes to prevent the risk of making the wrong decision driven by AI. And there are lots of gates that you should cross when you do those decisions. But eventually it's really about the use case.

So just think about what's your use case? What's the solution that's actually going to work? So yeah, going back to your example about the heart surgery, I've seen those dancing robot videos. They look very cool, but I'm not sure I'd let one of those things like operate on my heart. But on the other hand,

If I had a heart problem, then a pacemaker, which is a completely automated solution, is perfectly valid because it does actually work and it's going to work better than having someone like a human just punch you in the heart. Like, once a second. So yeah, pick the use case and see what the best option is there. Getting back to this idea of decision intelligence and making decisions using data, it seems like the hard part is scaling this. So I think every organization, you're going to make some decisions with data.

How do you be more consistent about it? Well, one of the challenges with analytics and data analytics is that it lives in silos. Because of the way analytics tools have been built, they were built to support either a specific function, a specific need, a specific group of people, maybe departments. And, you know, when you look at upper mid-market and enterprise companies, usually those big companies, they have everything under the sun. And

And it's not really built for consistency. It was built to solve a very anecdotal problem that I had. So one of the challenges is, you know, think about it. You have one CRM system, which CRM, by the way, used to be also like 20 different tools, my customer list and my Rolodex and my customer leads and my this and my that. And it slowly but surely consolidated into a singular solution, CRM, ERP, the

The same. And analytics is still fragmented. It's all over the place. And I think that what we're seeing today as a trend is that vendors are starting to figure out how to consolidate different aspects of your analytics stack into a similar wholesome experience from a technology standpoint.

So that's the first step to try and solve for inconsistency because if my data acquisition, my ETL tool is a separate entity and my dashboarding tool and my data discovery tool and my reporting tool and maybe my data science workbench is a different tool.

All of a sudden, each one of them is a whole entity. They don't necessarily talk to each other, but they need to be connected. It could be quite a challenge for companies to maintain the flow and consistency of analytics experience across those. So I think

One challenge that I think will definitely need to be resolved over the next few years, I think we as a company, that's what we believe will happen and that's why our infrastructure is built like that. It's a kind of a wholesome experience with multiple disciplines in the analytics space combined into a singular platform. But put us aside, I think that's where the market is headed.

into consolidation of your analytics and data stack into a platform. So that's one. Second, I think that the adoption of analytics across non-technical users is also very important. Usually you have the practitioners, the people that you mentioned earlier that are building those dashboards or building those insights, building those narratives for us.

There's IT, people that will provision and facilitate those solutions for us. But eventually the adoption rate is by the line of business, the people that eventually will need to use those systems. And I think that if I'm using data to make decisions, and I mean, let's say the marketing department,

and my colleagues will not do that, then it's a completely inconsistent experience. Again, we're back to people are eventually the hardest problem to fix. So I think the consistency is on the tech side, and it's also on the human behavior side. Absolutely. So I agree again. Yeah, the humans are generally hard on technology. But just on the technology side for a moment, it sounds like you're advocating for having a simpler business analytics stack using fewer tools.

I'm not so sure that it's simpler, but it's more consistent. So what is your analytics stack typically? It's few disciplines. It's the data acquisition. You acquired your whatever, your EDW or your data lake, and now you have data and you want to access it. So you access it with an ETL tool and you want to build out some models from it so you can actually start

exploring it. So you put on top of that ETL, now you build, you acquired a data discovery tool, and then you want to report the builder to push it out to your customers maybe. So now you have a reporting system and now you actually want to build all kinds of predictive analytics capabilities. So you bought your data science workbench system.

And each one of them is a different vendor. It's a different story. It's a different technology. They don't necessarily talk to each other. So I think one problem is super siloed

implementation. It's all over the place. It's disjointed. It's very difficult to maintain. In our space, it's called TCO, total cost of ownership. It's not just a license cost, but it's actually maintaining the entire infrastructure intact and actually operational. And usually it breaks. Why? Because it was never meant to be connected.

So that's one thing that I think we as a company are trying to address and solve for, but I think that's where the market is headed. The market is looking for not necessarily the best of breed for each one of those tools, but actually the best experience. And the best experience will come from

systems that will work, that will give you insights on the fly, that will connect to the right data in near real time or real time, that will be available for you, that will cater to any persona in the company and not just to the most advanced users. And then it goes on and on and on. So I think

Eventually, simplifying the experience and the ease of use and the democratization of access to those tools and systems is what's going to make analytics, you know, it's almost sometimes I think about it, I wake up with shivers at night that maybe a decade ago, analytics was nice to have. It wasn't like a must-have system. It was nice, fine, you know, build me a nice pie chart. It's not about the pie chart.

It's about building a consistent decision-making process that actually will help you compete in your industry, that will help you be successful, that will help you make the right moves in the market, that will help you build the right products, and this goes on and on. So I think, luckily today, analytics is a mission-critical application. You can't move anywhere without those kind of systems.

And so I guess moving to the people side of things. So since this is the hard part, I guess you want everyone in your organization to have some kind of data skills and maybe even business intelligence skills. So first of all, what do you think are the most important skills that you think people need to know like throughout your organization?

You need to be curious. You need to be able to ask questions. You need to be critical about yourself and your business. But those are characteristics. Not everybody has those. Some of us, you know, live differently. So I think that even if you don't have those ingredients in you, analytics helps you.

It helps you to be curious. It helps you explore. It helps you ask more questions. It helps you be much more critical about what's going on in your business. So you need to maybe at least keep an open mind that it's okay to ask questions. So that's one. Second, I think that analytics is data. And eventually, you need some capabilities around analytics.

the logic of what the underlying assets are built off. You know, what is a data lake and what is the difference between a data lake and a data warehouse and why it's important to use one or the other and maybe both. So I think having the basic understanding, at least for people

in the line of business, it will give you context of what is it that you're using. And then at least understanding the foundation of what is analytics and how do I connect to the data and ask questions and get some insight. So at least

I won't think that it's complete mumbo jumbo. Some of it actually does make sense. And maybe the last part is how deep am I going to be as a practitioner in the decision intelligence space? I actually think today, everybody needs to know how to ask the next question from a data and data analytics perspective. So if you're using a BI tool,

and you got something, your business is underperforming, you want to know why, you need to know how to ask the next question. Rail down, slice, dice, check your hierarchies, figure out the next layer of information, add more data into your data lake. Ask someone to pull your marketing campaigns, not just your sales performance data. So I think that you cannot not know those basics, at least.

Okay, so I really like the idea of asking questions being an essential skill. I guess as a podcaster, I'm biased. I ask questions all day. But do you think generative AI is then changing the kinds of questions you need to ask or the skills you need in order to be able to ask good questions? I think it opens the spectrum of questions you can ask. If before...

Analytics tools or BI solutions have functions, maybe a hundred functions, maybe a hundred thousand functions, but it's limited to the amount of functions that the vendor developed inside your BI tool. And that's what is available for you. I think the beauty of generative AI, or in our case, generative BI, is that you're using

your language. And then you can ask whatever you want to ask. You can build whatever context, you can add whatever context to your questions that you want to ask. So I think that it almost created a limitless

world of question asked. And in that perspective, I think that, you know, as much as it sparks your imagination, you can ask to connect things that might should not be correct. My business is exploding. How many people showed up in the office this morning? And maybe you'll see that there is a correlation between those two things and maybe not. And, you know, I think that

The beauty of using kind of LLMs to ask business questions is that it really can bring in lots of unexpected answers that are valuable.

Absolutely. So I love that it's kind of opened up the scope of questions you can ask. And it just seemed like imagination or creativity is now an increasingly important skill. Okay, so how is this changing the role of being a business analyst? Is the job different now with all these changes in the tooling? I think is AI going to take over different roles and functions in our businesses? And I think the answer is yes, it will. The agriculture revolution,

changed the way we plant and we harvest our crop. The industrial revolution changed people's occupations. The internet revolution changed the way we communicate and learn and build businesses. So I think that, yes, it is an everlasting changing story. And I think that we as humans and practitioners will need to learn how to adjust and adopt

to the fact that some of the mundane and some of the other jobs will be replaced by systems and AI or robots or whatever it is. But I also think that back to your question, is it going to change or replace the analyst? I think this is where I think it's going to be difficult for

AI to replace people that are curious, that are excited, that are interested, that will ask the non-trivial questions, that will bring their own

perspective into the mix. And I actually think this is the back to augmenting humans with machines. I do think that the successful data analyst people will be the ones that will take advantage of the ability to actually build out faster, smarter, better insights for their company, augmenting their own capabilities with what the systems can produce.

Okay, that's interesting. I suppose, yeah, I'm not sure whether AI can reliably discover the question behind the question and be able to generate insights on that. Okay, so beyond just individual data analyst or business analyst roles, do you think data teams are changing now because of changes in technology? I think that A, there's much more focus on data. And because of that,

They're becoming much more strategic and valuable for companies. So, you know, historically, maybe you had the data group only part of IT. Today, it's their own division, right? There is a chief data officer. There is a CDO now. Now, it's probably in the last decade or so, but I think it became super evident that data is a significant asset.

If it's a significant asset, your team that runs your data asset, your data estate, your analytics estate becomes your critical asset. So that's one, I think that it wasn't the case a decade ago.

I also think that you'll find the need for many more data scientists, and there is a shortage of them. So the world created citizen data scientists, which is basically people that maybe have less experience or are starting their journey in the data science world. And they're part of that group, that team.

So I think that new roles are emerging and kind of creating in the data group. And I think they're becoming super valuable and critical. They're probably the hardest to find.

Okay, so that's interesting. I like the idea of treating data as assets and you really think about what's the value there. Do you think there's anything that sort of traditionally worked for data teams that you think, okay, that's no longer the case, just stop doing this now? Yeah, I mean, I think there may be lots of things that have changed, but the fundamentals didn't change. The fundamentals are...

Keep the data, refine the data, acquire the data, distribute the data, use the data to run your business. Those things are maybe hasn't changed forever.

What changes is how it's done. And I think the how is, A, once it became an asset, then people treat it like an asset. And it's not just an afterthought. It's actually something that is super well architected, thought through. It's secured. It's governed. Sometimes it's a secret. Sometimes it's compliant.

Probably the things you hear the most when it comes to data is data breach, right? Every time there is a leak of some people's passwords, the world collapses. Imagine what happens if you're one of the credit card companies and your data has been breached. You basically lost your business. That's your business. Your business is that data. So I think the data as an asset changed everything.

Everything around it from human perspective, roles and responsibilities, the type of knowledge and know-how that you need to have in order to treat that universe eventually. To me, data today is everything.

Absolutely. And yes, certainly data breach is incredibly worrying, far too common. Cybersecurity is just hard. So are there any things you think that data people need to know about security? That is important. And you don't want to breach it. And you need to invest in it. I don't want to go into the technicality of it, but I think that because data became an asset,

People steal assets. They don't steal nonsense. They don't go after things that will not be valuable. They're going after valuable assets, which means that the same people that would like to steal your car, or maybe not the same people, but people will make their life experience to try and steal your data assets.

And the more your asset is important and critical, the more security gates you need to put on top of that. Obviously, we're not in the cyber space, but there is a reason why cyber is growing double digits year over year. It's because the amount of attacks that are being done on data today is probably growing more than the double digits that cyber is growing.

Okay, yeah. I like the analogy with the car because certainly you wouldn't just leave your car with the doors unlocked. But maybe that happens a bit too often with data. It's like, well, yeah, it's not protected well enough. All right, super. So just to wrap up, what are you most excited about in the world of BI? Yeah, I mean, listen, everything we talked about. For one, I'm super stoked about the generative BI side. And I think that it's going to...

There's been a gradual progress and progress in BI. I think generative BI is going to be a quantum leap. I actually believe that it's going to allow everybody to use insights to drive their business. If it was difficult and cumbersome and annoying and unexplainable, how you got there, what to do with it, I don't want to use those BI cranky tools, all of a sudden,

with those generative BI systems, you can actually give it to everybody. And I think once it's in the hands of everybody, everything changes. So to me, you know, Gen BI is probably the most exciting. It's a dramatic change in the analytics space. BI for everyone. It's a nice thought. Okay, wonderful. All right. Thank you for your time, Amri. Thank you, Rich. I appreciate it. Thanks for having me. It was super great. Thank you so much. Thank you.