Evolving the Enterprise

AI in Data: Unlocking Possibilities With Dawar Dedmari, Data Engineering & Analytics Leader at Meta

• SnapLogic • Season 3 • Episode 2

Share episode

In the ever-evolving landscape of data and artificial intelligence, there's a wealth of potential waiting to be unlocked. Join our host, Dayle Hall, as he embarks on a captivating multi-part conversation with Dawar Dedmari, an accomplished data expert who brings a wealth of experience from the tech industry. This podcast series delves deep into the intricate relationship between AI and data, offering insights that will intrigue data enthusiasts, inspire business leaders, and satisfy the curiosity of anyone interested in the transformative power of AI.

Throughout this episode, Dayle and Dawar explore a wide spectrum of topics, ranging from the fundamentals of AI and its practical applications to the nuances of data integration, the critical importance of ethics and privacy, and the promising future of AI in the realm of data analytics. It's a comprehensive exploration of AI's ever-expanding role in shaping how organizations collect, process, and leverage data to drive innovation and informed decision-making.

Discover how AI is revolutionizing data collection methods, simplifying complex processes, and enabling data-driven strategies that were once considered beyond reach. Dive into the nitty-gritty of data privacy and ethics, exploring the challenges and opportunities organizations face as they navigate the uncharted territory of ethical AI use. Learn about the transformative impact of AI on data visualization and how it's changing the game for businesses seeking to gain a competitive edge.

But this podcast isn't just about the technology—it's about the people behind it. You'll gain valuable insights into the minds of data professionals, uncovering their perspectives on AI's evolution, the challenges they face, and the exciting opportunities that lie ahead.

Sponsor
The Evolving the Enterprise podcast is brought to you by SnapLogic, the world’s first generative integration platform for faster, easier digital transformation. Whether you are automating business processes, democratizing data, or delivering digital products and services, SnapLogic enables you to simplify your technology stack and take your enterprise further. Join the Generative Integration movement at snaplogic.com.

Additional Resources

AI in Data: Unlocking Possibilities

Dayle Hall
Hi, and welcome to our podcast. I'm your host, Dayle Hall, CMO of SnapLogic. This podcast is designed to give all the organizations out there some insights and best practices on how to integrate, automate, and hopefully transform their enterprise.

We have a very special guest today who has played a pivotal, a massive role in leveraging that thing we've all heard about, generative AI, but also modern cloud technologies, understanding traditional data approaches to create core data engineering and analytics solutions at Amazon. And now he's at a very well-known organization called Meta at their Reality Labs. He has a proven track record of leading and scaling data teams. He’s got valuable insights on how generative AI and automation is revolutionizing the data analysis industry. Please welcome Dawar Dedmari, who's the data engineering and analytics leader at Meta. Welcome to the podcast.

Dawar Dedmari
Thank you so much, Dayle. Super excited to be here. And thanks for having me.

Dayle Hall
Absolutely, absolutely. Typically, what we try to do as we kick these off, give me a couple of minutes about you, how you ended up where you are at Meta, but also specifically around this area of data engineering and analytics. What drove you to this? What was the background? What got you to where you are today?

Dawar Dedmari
The short answer, I would say, is by accident. But again, I think to provide more context, I graduated, basically, my specialization was in electronics and chip design. That was my first job out of college. But that was also the time of the Great Recession, as the people who will be older in this podcast may remember, that we had back in 2008 and ‘09.

Dayle Hall
No, I don't remember that, Dawar.

Dawar Dedmari
As a result of that, the whole hardware and chip design industry fell into a downward spiral. And then I was looking for a new job. Data wasn't cool back then. We didn't have fancy titles. Data engineering didn't exist. Data scientists didn't exist. Most of the teams were either in economist teams within companies, and we were at the start of the data warehousing cycle and stuff.

I got to start in one of the data teams back. They were creating the first data warehouse stock in Oracle. Oracle Cloud was the new thing that was coming to the cloud. I started my data journey and career right there. Business intelligence has just come to the enterprise back then. We didn't have Tableau, for example. Oracle BI was in its initial days and stuff. I grew up in a world where teams were still figuring out what does data engineering mean, the old Kimball traditional data warehousing architectures most of the time, and that's where I got my data chops.

But then data became the new oil. I got a chance to work in a bunch of teams at Amazon as they were going through the whole journey of how do we really leverage data and monetize it, how do we transform it for customer service. That was also the time when the whole shift from data warehousing to big data was happening along the way. Data infrastructure teams were coming along. Spark became a big thing and stuff. I worked on a bunch of tools within Amazon, and then a couple of years in their devices team. They were really different products. I'm sure maybe some of the people here have used it. That was generating a tremendous amount of data coming in.

And then the question was, how do we build frameworks around data? How do we use it to offer personalization, do things like customer segmentation and stuff like that? The surety of data teams was happening as I was working for Amazon. I would say I was lucky to be in places, which was all about data. We got to build a ton of those solutions at Amazon for that.

And then at the end of last year, as we were coming out of the pandemic, most of the people were not aware of Reality Lab. That's a division of Meta, which works in our AR, VR, XR space. Meta was looking to go from [inaudible] labs into more of like, hey, how do we scale it up as we grow our customer base. It was a great opportunity with great leadership there. And I was super excited to work in a new place, which was small enough and then wants to grow big enough. That's what we had done at Amazon, start small enough and then scale it up. So it was a chance to build again, and I love that kind of building new stuff and solutions with data. So I was super excited to join at that point in time.

I do want to say that whatever I say in this podcast, this is me expressing my personal views. Those in no way reflect Meta’s position on anything. This is just me and my own personal views.

Dayle Hall
That's a good disclaimer to start with. Again, I think Meta is a very popular and well-known company. So I appreciate you saying that.

Look, I think you've definitely seen, whether it was accidental, or one way or another, you've seen probably a lot of innovation. And obviously, recently, with the advent of generative AI technology with some of these companies coming out, and obviously now that a lot of big players are getting in the game, which we expect that, and you've seen a lot of technology development, but is there something specific over the last- I'm just going to say six months because everything seems to be moving faster than I have ever seen. And I've been in tech and the marketing tech world for 25 years, too. But is there something you've seen specifically recently that just gets you excited, something that's like, wow, I never thought we'd see this? What have you seen that has really piqued your interest?

Dawar Dedmari
Oh, absolutely. I think the last 6 to 12 months have been a very exciting space for people who have been in this domain. One thing which was a change for us was there were people working in this for the last four to five years. GPT is not a new thing, for example, right? So there were smaller teams, research teams, and organizations who are already working on this stuff.

But I think the key thing that changed with the release of ChatGPT was this suddenly burst into the public space. Because the way they released it was a simple interface, open a website, just type something into it. I think it brought AI into every single home, I would say. Everyone was trying to play around with it. I think it caught a light bulb moment in the industry, which was like, hey, AI is not that far that we thought, like years into the future in [research]. It's here and now.

And then the question became, what's our AI strategy? I'm sure a lot of leaders and a lot of data teams were suddenly asking that question to the people, hey, what's our AI strategy? Do we have anything, or any project, or working on it? And if not, what can we do now? How do we get something on the plate? What can we implement?

So I think like a bamboo, it's underground for five years, and then suddenly shoots up in six weeks. I think the industry had that moment in the last 6 to 10 months. We are seeing most of that- again, I don't want to use the word hype cycle came in, but I think it also sheds light, and there's a lot of good things happening there.

Dayle Hall
I love that analogy, actually, being the bamboo of the latest development because it does feel like the second it became something that- I don't want to say regular people, but a marketer could use the interface, they made it really easy for someone to create content, at least in my perspective. Then all of a sudden, whatever has been developed for the previous four or five years, it just took off. It's incredible.

In your areas specifically around data analysis, what are some things that you've seen using this kind of technology, either through that simple interface through LLM or just in general AI around data analysis? Where is it being used in your line of work?

Dawar Dedmari
I do want to call out that most of the products in this space are in their initial versions, or some of them are in previews and beta views and stuff. A lot of the things that we are seeing happening right now are primarily early POCs or beta releases that are happening. We haven't seen a lot of drastic change in the adoption at the enterprise level, if you want to ask about it. When I say some of the things that I'm going to say, we have to caveat it with the sense that these are early days, just getting started.

But we already see, for example, one of the biggest areas we are seeing AI being used is synthetic data. One of the big blockers for building a bunch of generative AI systems has been the lack of data, which may sound paradoxical because we usually say there is so much of data. We are getting into 0 bytes now, right? But good quality data that you could train around publicly or privately within an organization has always been a challenge. But with the advent of a lot of synthetic data, that kind of goes away. So you could do so much more in different flavors because of synthetic data coming in. I definitely see that happening and helping the data teams.

Also, a lot of the work around data cleanliness, collection and processing, that had a lot of manual input. There are solutions being built specifically to address that space. In my view, I think a lot of that is just going to get a huge adoption because that is just work that AI is better than a human.

And then also, I think one of the other big things that will transform and we are seeing the starting shoots of that is the democratization of data, as I call it, or democratization of analytics. In the old world, we had data teams. We were like the protectors of the enterprise's data. And if you, as a normal user, wanted to get access to data, you go and talk to the data team first, and then they give you their time and attention and a bunch of stuff. I think a lot of that model is going to change with the new tools that we are getting because a lot of enterprise data is going to connect into things like conversational data assistance and stuff, or even help people who don't, for example, know SQL very well get access to the data that they want in order to do their job. I think that's just going to happen over the period of time.

Dayle Hall
Yeah, it's funny you say that. One of the things that the company that I work for, SnapLogic, had done is exactly that. We have an interface where instead of understanding coding and knowing what apps and what data sources do snap together, you can type in what you want to do, and it will build that data pipeline for you. I think that is a great example of democratizing data and it will become more available.

What do you see as the potential risk in that? One of the other questions I had as you were talking about all these things that we get with data and collecting, is the problem going to be collecting? Is it securing? Is it cleanliness? Is it processing? Even with these models, what is going to be the big challenge?

Dawar Dedmari
I think there are challenges in each step of the way that you're describing it. For example, data collection, we do have risks associated with the bias in AI, the hallucination that they have, and how does it impact the data that we collect. For example, if you are trying to impute data, a lot of AI systems now are doing real-time imputation where data collection happens, but if there are errors in the data or missing data values, the AI will impute it. But what does that impute? It will impute based on the inherent biases that it may be trained on. So you could have some sort of data pollution in that sense happening, which reinforces the biases that the AI already had in that sense. So we do have risks in data collection.

And then along the way, if you see on the other side when we are producing output and insights, the biggest risks- or not a risk, but the challenge I see is AI may lack context. It's really good at munching numbers and giving up- for example, we are seeing a bunch of tools do summary insights. Most popular tools are coming up with that. But sometimes what it misses is context behind it. And that is hard to do with AI at this point in time.

Let's say the AI produces a summary result, and it goes to an exec leader. And he has to make a decision based on that. Then I have to talk about how much trust do we do in that. I see those kinds of challenges because we are in the early days yet. So we don't know how it's going to pan out. But for example, if my AI spits out a summary, hey, we have this error in our sales and stuff, and then a leader depicts a decision based on that. How much does he need to trust that data and insight? If we say, has a human checklist. I don't feel this passes the smell test and that. I see we have challenges in every step along the way that we're still trying to get around.

Dayle Hall
It's a good point, which is slightly off topic, just based on what you said. You said AI is going to be better at some things than a human, like some of the collection. We have the bias side. Then you used the example, then how does an exec use that data and can he fully trust it. I've talked to a number of people on these podcasts around this concept of will AI take people's jobs. We've actually been talking about that for years, right? That isn't new. But it feels like it's coming back to top of mind.

How do you think about- is it going to take people's jobs? Or will they be able to do new and creative things? Who sits between the AI output and that exec that is going to have to make that decision? Who's the gatekeeper to make sure that the exec doesn't read something that one of the GPT tools create, and I'm like, I'm going to change the entire business direction of the company? How do we manage that?

Dawar Dedmari
I'm an optimist.

Dayle Hall
Okay, good. I like that.

Dawar Dedmari
I'm an optimist. So when people paint these worst-case scenarios of AI taking over all our jobs and stuff, my short answer to that is I don't think that's going to happen. But are we going to see some short-term disruption? And what activities do you do in your job? Yes. And we're going to say that in the short term, that over the next few years, for sure.

The way I look at AI is like another tool in your belt. The tool is complementary. A lot of the stuff, for example, that the teams or data analysts or business analysts in the team do right now, the AI can do it much better. A human doesn't have to do those repetitive tasks again and again. For example, creating manual reports. A lot of data teams are just producing reports every single week for a weekly business review, a monthly business review, a quarterly, whatever you have. Is that the best use of their time? Probably not. There is so much more that they can do beyond producing numbers. I think the AI can produce the numbers, but then the human comes in and adds context, adds narrative around that number to make a more powerful business pitch for the exec, like, hey, here's a number, here is what it means, and here's what we can do about it.

My stand on AI is- and I think Ali Dalloul, I think he’s the VP at Microsoft as your AI platform, he said that his prediction is that over the next year, everyone will have an AI assistant helping them with their job, not taking away their job. I want to say that every new technological advancement, and AI is bringing one of those, people will have to reskill, upskill, do different sorts of activities than they do currently. But I don't see overall- I think we're going to get new jobs. Again, I say this, I was having a conversation last week, we didn't have prompt engineers last year. That job did not exist. People are hiring prompt engineers and stuff. AI will take away some aspects of our jobs, for sure. It will also create new jobs, for sure. And it will change the nature of many of the jobs that data professionals are doing right now.

Dayle Hall
Yeah. I like what you said, actually, and I haven't heard it described this way, which is people will see a short-term disruption of activities in their job. And it will be- I've heard it called [Pilar], I've heard the assistant, but I think that's right. And I think if people look at it more of a, this will disrupt what I'm doing, I will have to do some things differently. But it should also give me scale. It should also allow me to do something maybe that I haven't been able to get to because I'm too busy running reports every week, every month, every quarter. So I like that perspective.

I want to just come back to something you said. We touched on it a little bit, which is around ethical AI and capturing of data. Let's talk a little bit about the specific privacy. You, in your role, how do you think about potentially managing bias? And how do you think about some of those privacy concerns? Should enterprises think company-wide about these things? Should the owners of the data be thinking about it? Because this area is still, as you said, we're very early on this journey. And I think ethics and privacy is going to be a big thing moving forward. So how do you think that we should address things like that?

Dawar Dedmari
Oh, absolutely. Working in data for the last 15 years, I am a huge proponent of data privacy. So let me start from that position.

Dayle Hall
In your role, I’m glad that's what you think.

Dawar Dedmari
Yeah. Because I'm a customer in most of the modern platforms that I use. So I want everyone to care about what they do with my data. That's my starting position when I look at data privacy. Would I want my data not to be handled the way I would handle someone else's? Yes, absolutely, sure. So I start from that position.

And I think there's a different role every single entity has to play. For example, organizations need to have end-to-end data privacy right from the get-go of when they're designing products and stuff. Privacy shouldn't be an afterthought. Privacy shouldn't be just because we have regulations to adhere to. I think privacy is and should be a key input into product design. How do we make sure that the customer data and privacy is respected and the product design?

I also believe there is a role for legislation and monitoring bodies to play here, at least the minimum bar and standard that all organizations should adhere to, at least. I think organizations should go above and beyond that. But there should be a minimum set bar of how to handle customer data. And again, I think we have seen many countries take a stand on this in the last 5 to 10 years. Regulations and laws have been passed and are passing. Every one year, some other country passes a law on that. So I definitely feel that should be there as well.

I think the responsibility then also lies with a lot of our customers as well. Because one of the key expectations I've seen is- and again, this is a meme on the internet, but they keep saying if you don't know what the product is, then you are the product. That's true to a large extent. So we need to be privacy aware as well. Organizations need to make it easier for customers to know what data is being collected and what they are doing about it. But a customer should also then use those tools to see what's happening with my data and hold organizations accountable. Yeah, I think I am more privacy aware since I've worked in this field. But some of the other times, I think people don't really take it as seriously as I would like them to take it.

Dayle Hall
Yeah. I think what's interesting you just said is who should be responsible for that. So customers- I think that's actually interesting. No one's ever expressed it that way, which is customers should also be understanding what is being collected. Because I think a lot of enterprises- look, you've been at Amazon, you've been at Meta, we've both been at Oracle. We've seen a lot of these companies get- I don't want to say attacked, but they're definitely hit hard around understanding the consumer, treating people with respect, the bias in there. So I think enterprises clearly have a role.

But then you also mentioned groups. You mentioned other groups. There's a guy I did a podcast with called Steve Nori that starts something called AI for diversity. He’s actually based in Australia. And there's a couple of others that are creating groups of us, could be an enterprise, could be consumers, could be my dad for that matter. But bringing people together to understand, how should we be involved in helping setting some of these rules? Because I think I know there's a lot of blame that goes on corporations. But I also think that kind of absolves other people from taking any responsibility. Oh, you run this tool, it's your fault, you should own it. And I think that is a good perspective that we can all be responsible for some of this.

Dawar Dedmari
Yeah, absolutely. The way I approach it is like, if the people don't get involved when the time for creating frameworks around privacy or laws around privacy happen, then we are doing a disservice to ourselves, right? Then we will always be more in a reactive stage like, hey, the law already passed, or the company in x and y, and so now I'm shouting about it. But if we are constantly involved in that conversation, we can actually drive the conversation towards the outcomes that we think are the best words for us as customers.

Dayle Hall
Yeah, I like that. I like that a lot. I think, again, we're definitely on the start of the journey. I think we're definitely going to see more people involved. We just talked about having kids. You have young kids. My kids are a little bit older, but I see some of the things that- my daughter is 16, going to be 16. This is going to be squarely in her as she's growing up, the things that she has to deal with. So I want her to understand what's out there with these kinds of tools. I talk to her all the time about the things that she's signing up for online because I don't think she fully understands. But that's my responsibility. It's also her responsibility to do that. And I think that's going to become a bigger topic as we go forward.

Let's move forward to talk a little bit about the visualization of data. Is there anything that you're working with around, is AI helping to not just- we talked about this, not just the capturing and the storage, but how is AI potentially helping how we visualize and potentially use data? What are you seeing? What are you working in your environment with?

Dawar Dedmari
Yeah. Again, not only within Meta, but beyond in the industry as well. We are seeing the start of those new solutions come in all the way from- for example, I've seen no-code dashboards become a thing very recently. I think three to four months back, I saw many companies release solutions where there is no BI developer or a data engineer creating a dashboard. Like you can introduce data to your AI agent, even without input. I've seen even extremes of zero input where you are, hey, here's some data, visualize it for me, and it will auto generate the whole dashboard for you. Or you could give it some of the caveats and requirements like, hey, here is what we are trying to do, can you create a visualization for it? We have seen AI do that. So we definitely are seeing AI get into the space of creating visualizations and enterprise dashboards and stuff. That's definitely happening now at this point in time.

One of the good things that I've seen happen is it gives you- it's like a writer's block. If you don't know all the permutations and combinations of presenting this data, the AI assistants can really help you on like, hey, what's the best way to present this data? Is it good on a line chart? Should I create a donut around that? Or whatever it is, right? It can create a lot of variations for you very quickly. You don't have to waste a lot of time prototyping stuff, going back and forth on that.

I think a lot of AI power is going to make our lives easier in terms of visualization because it can create all those variations for you, it can create the prototypes up and running very quickly for you. And then you can have a more meaningful conversation with your customers of like, hey, what's the best way to do this among all the ways that- I have come up with the help of my AI and support system. I definitely see that happening more and more.

Dayle Hall
Yeah. I think that's really interesting. Look, I remember, when you put data in Excel and you get to pick, is this a bar chart, is this a pie chart, that's basic stuff. If we're going to be able to have tools that say, based on the data that they're seeing, present it in a different way, visualize it in a different way, suggest things, that's pretty powerful. We talked about execs making decisions about something that can help present it in the right way that they can make faster decisions. Because a lot of the things that- again, we talk about this at SnapLogic, but a lot of things is how can you use this, not just to meet your daily life better, but can you make business decisions quicker? Can you help customers get more out of the product or be more productive? That's what we're trying to do.

I wonder if we're going to see that kind of visualization of data or the usage of data. Do you think we'll see more of that from our own internal perspective so we can make decisions quicker? Or will we use it more to help our customers be more productive within our products? Where do you think the sweet spot is?

Dawar Dedmari
I think both things are happening right now. One of the things that's happening is called audience-based visualization. You take the same dataset. You may have a whole spectrum of customers from really deep dive account managers sort of stuff, like operational teams, all the way to the CEO, CFO and stuff. The data remains the same, but they want to see different aggregations, different cuts of it from the top to the bottom right. The operational teams are interested in doing the day-to-day job out of it. The execs are probably looking for high-level strategic decisions that they want to make. And there is a whole spectrum in between.

I think one of the ways in which AI is powerfully going to do this is to take a data model in, and then it'll become audience aware, as it's called. So create visualizations in many permutations, depending on the audience, who is looking at it, and then help them do their jobs better. I think that's really powerful, because that will take a whole data team to do it. And now you could do that.

The time to market is going to reduce drastically when you have AI doing this stuff for you, because otherwise it will take you a bunch of engineers working on a maybe three-month long project to deliver it. But with AI, you could basically do it in two weeks’ sprint and then iterate on it. I definitely see that happening a lot.

Dayle Hall
Do you think other specific organizations that are leading the charge on all the things we talked about, on the capturing or using maybe it's the visualization and actually using it more in their approach with customers? Is there anyone within the organization- obviously, you're very engaged in a lot of this because it's right in your wheelhouse around the data. But is there a specific line of business that are looking at this and saying, oh, we can really accelerate what we're doing with this? Or as we've talked about, because it's still early in the journey, everyone's putting their toe in the water, but no one's gone for a swim yet?

Dawar Dedmari
It depends on the maturity of the organization, to be honest. The people who can use these tools now are teams and OGs, whose, for example, their metric frameworks, their data frameworks are already mature. Because then I can come in and roll out quickly solutions for them. So they can take advantage on it today.

But then there are organizations whose underlying data infrastructure or metrics are not that well defined, or they are in that journey of getting up there. For them, AI is not going to be effective tomorrow or today. They still have to do the journey. You have to meet AI halfway.  AI is just not a magic wand that- let's say a company has no good metadata, no nothing, no metrics set up. And I want to bring AI into that company. What's AI going to do? Struggle, right?

But let's say we bring AI to a company, which has a very mature data infrastructure already in place, AI can do almost magic for them. I was working with a friend last week. And this is something very powerful, and people who have Code Interpreter and some of the Open Access already. A product manager was writing a doc justifying a business strategy. They reached a point, they were like they wanted a data point in the doc, and they could replace it with a xx and ask AI to fill that in right there as they were writing their thought process out. That is pretty powerful.

In the old world, they would have to reach to a data team, ask for data, maybe get that data back in a week, or whatever that timeframe is. And then by that time, they've already lost their thought chain and stuff. And it's already all in stuff. In my worldview, the teams which will benefit most from Ai  are teams which already have a lot of data maturity right now, because they can get to start to use it out of the box. But for the rest of us who are still in the journey of creating that data maturity in their organizations, AI just doesn't solve things tomorrow. It can solve a lot of stuff, but it's not going to be the same effective as it would be in the maturity of the organization.

Dayle Hall
That's really interesting, and I'll tell you why. When I do these podcasts, I'm always thinking about someone's out there listening, what is the nugget that they're going to take? I remember that. But that's really interesting because then, whoever is involved in these kinds of activities, bringing in AI, they should also be able to set the expectations with executive teams who expect we're going to bring in an AI tool, and it's going to change our world.

But like you said, if they don't have the data, the metric framework that's mature enough to be able to use it, it's really important that you have that conversation up front saying, these are the architecture fundamentals for a data we have to put in place, then we'll be able to do this. But we won't be able to just drop in this generative AI tool and we're off to the races.

Dawar Dedmari
Nope. In fact, I think you could cause chaos in such an environment, honestly. Because let's say I have 50 different measurements of sales. I'm just making that example up. And then someone asks a question of like, hey, how are our sales last week? Imagine going to exec meeting with five different people bringing in five different numbers without by AI, because they didn't have a framework below the AI working on it. So all it's going to cause is more distrust in the data rather than helping them out.

Dayle Hall
Yeah. No, I like that. Again, I think one of the things that- whilst we talk a lot about the art of the possible and what this is going to be able to do, I think there are some underlying structures for data architecture within the organization that have to be in place. Who's the best person to talk to the execs about that? Is it the data team to make it clear what needs to happen? Is it the people that are buying the technology? Is it the lines of business that are looking to use the data? Who runs that? And how do you get that alignment across the enterprise to make sure everyone understands where it is?

Dawar Dedmari
I think it should start with the data team in an organization, because let's say a business organization wants to bring in AI and do stuff, they probably don't have an idea of the under-the-hoods data infrastructure and maturity that they have. If it's just that organization having that conversation, it's not an effective conversation. I think, overall, I would say the data team can lead it, but they also need support from execs and then the business groups about, hey, what are we trying to do here? Because then they can say, hey, if you're trying to do x, here's where we are with x in terms of a data platform and infrastructure maturity. So if you bring in AI, it is what's realistically possible in a journey. And then we could move on to this and that. I think it's mostly driven, it should be driven by data teams because they know the data best. But then they also need input from execs and stakeholders around them to make sure that they're God willing that conversation.

Dayle Hall
Yeah. And I think that's one of the things that I think is interesting is, how do you really put these tools into your business processes. And we talked a lot about having the underlying infrastructure. The organizations you've been at, you've been part of large projects, and change management through these processes is usually overlooked until something goes wrong. And then you do a root cause analysis, and then you go, okay, next time we do this, we have to manage change, as well as looking at who should be responsible to drive it, as well as setting expectations.

How do you think about some of these technologies and potential change management challenges? There's obviously benefits, but what are the challenges of saying, okay, this is going to be a massive shift? How do you manage that? Do you create a board or a council? What are the things that you think would be most effective?

Dawar Dedmari
I've seen many versions of this happen, like some people create data governance councils, or data strategy, or data architecture councils and stuff. The committee approach has its limitations, if you ask me, honestly, especially if you're trying to move fast. I would approach it in a start-up kind of a mentality. If you're a large enterprise organization, you can't just tear down the old world and say, hey, we're going to start using a new world from tomorrow. It's a huge disruption.

The way I would approach it is you create a smaller pilot group within the data organization and give them very specific goals. Don't make it a pie in the sky, like a dream. No. Give them a very specific business process and say, hey, here's a business process right now, which has a ton of issues. Here are some of those issues. Can we bring an AI into this process and improve it? This sets them up for success. Many times what happens is we see organizations, they bite so large, it's hard to chew sort of stuff. And that sets up data teams for failure. I will see a more milestone-based approach where they're like, hey, let's bring it in on something which we know it can help. We have good goals and output measurement to see if it's success or not. And then we can iterate around it before we release it more in the broader organization.

Dayle Hall
Yeah. I like that approach. I think one of the things is we talked about ownership, managing change. And there's this term that has been around, particularly in the technology industry, which is low-code, no-code, of which you can argue generative AI and those kinds of LLM is no code. It allows us to interface. Do you think that the advent of generative AI and that simple interface, does that mean the things that people have been talking about low-code solutions, they're gone?

I'm not saying coders won't have a job anymore because there's a lot of people that still enjoy doing that kind of thing. But are we on a path now where that is just going to become commoditized? And we talked about prompt engineers earlier, which is it's all about the right question to ask rather than knowing how it works in the background. Because I think that low-code, no-code discussion, particularly in software tech, and we're a software company, and you're a tech company too, that seems like it's been going on for a long time. I just wonder where you feel like- think about where the industry might go now we have this generative AI technology.

Dawar Dedmari
My stance on this is, again, like you said, we have been having this back and forth on low-code or no-code for a while now. Every few years, we have some new tool and stuff launch or an announcement happen. And people raise their hand and say, this is the end of code and stuff. No, it's not.

The way I see it is what's going to happen is two things. First of all, a lot of the LLM-based generative AI, let's talk about that, at least what's available right now, a lot of enterprises are going to have a hard time connecting it back to their own data without any code in place. I can't ask business questions to ChatGPT about my own organization. It just doesn’t have that answer or data.

And again, enterprises are going to have a hard time sharing their data, business data, what I mean, with public entities, like some of the foundational model companies that we have right now. It's not going to happen that way. We need to just realize that it's funny asking, hey, who's the president or the oldest president of the United States versus hey, how much sales in my department do last? That's just not there.

I think the way this will evolve is there'll be a handful of companies doing foundational models. There will be a large interface layer of companies in between, which will enable enterprises on the right side, if you want to visualize it, to use foundational models but then combine it with their own data to get answers which are relevant to them. And all of this would require code in place. This is not going to happen with no code. There has to be integration built between foundational layers, integration layers in between enterprise and their data, to get to the output.

For a large customer base that we’ll have, their lives will become simpler. We don't want, for example, everyone in the company to know SQL down the line, yes. But then the interface would be more conversational or integrated in whatever business tool that they are using. But then what powers that would be an integration layer in between, which will mostly be code. I think that's the way I see it happen.

Again, when Tableau released, a lot of people did bucket Tableau into the no-code sort of a place. And yeah, you can basically do simple dashboards with that. But then eventually, teams realize that to power a lot of things behind it, you still have people to call and do calculations and a bunch of stuff. I don't think coders are going to be out of a job very soon. They may have a little bit of a different job profile than what they do right now, but I do feel the value to a lot of customers who had an entry barrier because of the code in place will get reduced over the next few years. And I'm all for it. The more people that use data, it's good.

Dayle Hall
Interesting. I think that's a great place to stop the podcast, because you just said one of the key areas, that integration piece between, that's exactly what we're trying to do with our company. It's definitely going to be a fun time. Dawar, I really appreciate the time. This feels like we're just- I wish we were having a beer, or this is a chat that we would have over dinner. I owe you dinner for being part of this podcast.

Last question as we close this out. I asked you at the start what you've been excited about seeing so far. If you could look into the future, 12 months later, things go so quick anyway, 12 months, two years, is there something specific, or something that you can see happening, or something that you're excited about, or something on the horizon and you would say, yeah, that would be really cool if that came to fruition, something that gets you going?

Dawar Dedmari
Yeah, absolutely. First of all, I think we will be done with the hype around this thing. And we'll get to the real benefits by then, for sure. That definitely excites me. I think a lot of the initial steps that we're seeing those integrations released right now in betas and stuff, they would actually hit production by then. Microsoft, for example, will definitely be fully integrated in their 360 with AI. A lot of people who use Excel, Word, and PPT day in, day out in their jobs will give them a thumbs up. I am looking forward to it as well. I don't want to create PPTs and stupid Word docs and stuff. And 30%, 40% of my job is sometimes all of that. So I'm looking for increases in personal productivity, honestly, where AI does a bunch of stuff on my task list that I don't want to get to. I'm super excited about that.

Dayle Hall
I love that. And again, I'm going to close with the quote that you said, which is short-term disruption of activities in your job, but it's going to have a lot of long-term benefits and be able to get to some of the stuff that you can't get to.

Dawar Dedmari
Yeah. I think the only thing that I would caveat it with is embrace the new wave of technology that AI is going to unleash. If you don't, then you're not going to end up in a good place, let's say, three or four years, five years down the line, because people around you will. And if you're an early adopter, things will work out great for you. Otherwise, you have to play that catch-up game. So I would say,  embrace it, try it, enjoy it, and love it. It's going to help you.

Dayle Hall
It's good. Dawar, thank you so much for being on the podcast. It's been a pleasure.

Dawar Dedmari
Yeah, it's been a pleasure as well. Take care.

Dayle Hall
Thanks, everyone, for joining us on this episode, and we'll see you on the next one.