The Inflection Point of Generative AI, Analytics Role in AI and AtScale + Dataiku with Jepson Taylor

Data-Driven Podcast

Listen to Jepson Taylor and discover the transformative power of generative AI and its impact on the future of artificial intelligence. Gain insights into the role of analytics leaders in driving AI adoption and overcoming challenges in legacy systems and data modernization. Explore strategies for success and understand the benefits of integrating platforms like Dataiku and AtScale in accelerating data transformation and delivering actionable insights across industries.

See All Podcasts Button Arrow
Quote icon

You can think of AI as being a catalyst. So it allows you to do things faster and that’s good, but it can also be bad. It can allow you to do subpar work faster. So there was that example in the news of that lawyer actually bringing some claim, some referenceable claims into the courtroom that were hallux they were made up. Shame on them for not fact checking what it was producing, but just like you can use these types of technologies to do, to check your grammar, you can also use them to check for exploits opportunities.

I love the startup journey and I hope more people attempt it. One of the things I tell people is when you go to do a startup, you need to chase an idea big enough that you’re willing to fail. And then I define failure. Failure is foreclosed into your home, bankrupt max out credit cards like having to ask friends and family for favors. And when you say that to college kids, you kind of get, you see this reaction on their faces, like, oh, I don’t know. You see the hesitation and then I go back to big enough and I say, and that’s why it’s so important that the idea is big enough for who, not for me, not for your professor, not for your parents.

Transcript

Dave Mariani: Hi everyone, and welcome to another episode of AtScale’s Data-driven podcast. Today I’m really excited to have a special guest, Jepson Taylor, Jepson, who is the chief AI strategist at Dataiku. a friend of AtScale and an AtScale family, in Boston. Really an a, someone in AI who you need to know and you need to read, and read about, because he has got a lot to say. So, Jepson, welcome to the podcast. 

Jepson Taylor: Thanks for having me, Dave. I’m excited to be here. 

Dave Mariani: Yeah, I think you, you have, you have, just such an amazing background. I would love it if you could just, we could just open up this podcast with you telling the listeners a little bit about your, you yourself, your, your path into data analytics and ai. 

Jepson Taylor: Yeah. Well, I, I think, I think I should have gone into computer science sooner. So in college, when I was introduced to computer science, it was kind of the stereotypical, oh, these are the people that make computer games. And I didn’t have an interest in doing that at the time. So I studied chemical engineering, but I was introduced to programming through web design, and that turned into computer vision and high performance computing. And I, I love high performance computing, like e even today, like if you said, Ben, do you wanna go to the Supercomputing Convention Yes. Like that would, it’s Disneyland for me. So I worked in semiconductor at Intel and Micron for five years. I worked as a quant at a high performance computing hedge fund trading on the news in 2012. And then I was the chief data scientist at HireVue, which was HR Tech for four years. And then I built and sold a deep learning company to DataRobot a couple years ago. So I, so I, so AI is everything I do now. You and I have joked in the past, would you rather have better data or a better model And as, as quickly as I could, tap the button for better mo better data, not better, not a better model for people in industry. They desperately want better data. I I almost said the wrong thing. I almost said I want a better model. 

Dave Mariani: . Yeah, I love it. We’ve always talked about that, that, you know, it’s garbage and garbage out. You know, that’s kind of where it goes. But, but you know, it’s like, I didn’t realize that you, you, you studied as a chemical engineer. and it’s just so interesting that how many, how few people who end up in technical roles like we’re in now actually, you know, studied that I was an economics major. So, so, and nothing to do with it. It was my hobby. but, when it came to, like, how’d you go from your interest in, in scale out supercomputing, I guess in AI was using for, from the forefront GPUs, which is sort of a form of that, but what was that How’d you, how’d you get into, how’d you get into ai Like what, what was the, what was the path there 

Jepson Taylor: I think it started with robotic process automation. So when I was at Intel and Micron, a lot of the steps that I would have to do for work, they seemed mindless. And so at the time I thought, well, there’s gotta be a way to connect software to do, rather than doing A, B, C, D and e, I just want to do A to E while I’m at lunch. And so I was working on writing code that’d get access to databases. P there’d be an excursion, which is when some product is unprocessed, it’s very expensive, you need to do some analysis. So building tools that would automate that analysis. So that’s kind of where it started. It’s not really machine learning, it’s just trying to, I like to joke that I’ve been trying to automate myself out of a job for a very long time, and that’s hard to do. So if, and I would encourage everyone listening to try to do that. Find part of your, your job that you’re overqualified for or you’re doing some mindless activity and try to try to connect the dots so that that’s where it started. Was that then hedge fund, that was the baptism. 

Dave Mariani: Yeah, I bet. Man. You know, it’s, it’s interesting you talk about that cuz I kind of see it as like, you want the right amount of laziness, right Laziness. And that I, I don’t want to keep doing these repetitive tasks, right So you’re gonna write some automation to get yourself out of some work so you can do fun, more fun things. but it is like, there’s, you know, you never wanna say be lazy, but you know, you, you, you don’t wanna put up with, you know, you, you definitely don’t wanna put up with repetitive work, which you see a lot of people doing out there. which is really, you know, so you and I are the same mind on that for sure. but, but let’s get to the, let’s get to the, the, the big question of the day. 

Dave Mariani: you know, 2023 looks like to be a, a real inflection point for ai. you know, and, and, and until G B T came onto the scene, you know, everything was sort of focused on, on, on how we can build models and how we can automate the, the work of the data scientists to build those models. And, and things seem to completely have changed. Ben, so what happened What’s the difference Why all of a sudden Yeah. is, is AI in everybody’s world Whereas, you know, a a just a few months ago, it was a, a niche really, and, you know, a niche and, and a technical niche of that. 

Jepson Taylor: So I don’t wanna get any credit for pretending like I saw this coming. I, I actually wanna do the opposite. So I’m gonna go back before chat. C P t and before chat C p t showed up, I was very negative when it came to deep learning. I, I built and sold a deep learning company. I know something about it. I felt like the technology was brittle. I felt like we needed something brand new. Hinton’s working on capsule networks and these other, I I felt like we’re gonna have to reinvent something from the ground up, in order for us to go to A G R A G I, which is artificial general intelligence. So I, I felt like deep learning won’t take us there. And then when chat G P T showed up, I was shocked, completely shocked. Mm-hmm. because this re to me it represented a crack in the dam that this will absolutely take us all the way to agi the utilities through the roof. 

Jepson Taylor: And it’s so funny cuz when chat G p T came out, there’s a lot of negative press around it can’t do this, it can’t do that. Mm-hmm. it can’t do this. And I, I found that at the time I found that pretty annoying that you’re completely missing the point. Mm-hmm. , it’s not about what this does today. It’s about this is gonna be a thousand times better very, very quickly. And sure enough, we saw GPT four comes online, it’s significantly better, like night and day better. And it’s only going to keep getting better. So, so why this is such a big deal, I think before ai, the urgency wasn’t there with ai. Like some organizations had it, some organizations were making money using it, and really only the biggest companies leveraged it mm-hmm. . But now any, everyone that’s listening can justify an open AI subscription, a paid subscription. It doesn’t matter what your job title is, you can use it every day for an hour doing something summarizing writing code. so I, so I, I think that’s what we’re dealing with this year, is the utilities through the roof. It’s, it writes 80% of my code. And the fact that it can summarize things the way that it does, it’s, yeah. It’s, the units of work are through the roof for humans, 

Dave Mariani: You know So Ben, so Ben, you, you mentioned that, you know, do a deep learning. Deep learning was always sort of like, oh yeah, yeah. Those guys are trying to do deep learning. It was always sort of off to the side and we’re always, versus we’re focused on, you know, what, what we think of traditional, what I think of traditional, ai, which is, you know, all data science, data scientists, building models, for a specific task. What, what, what was the, what’s the difference with G B T cuz G B T was out there Is it just that chap G B T put an, an interface on top of it, it exposed it to more people and, and showed its utility Or is there, is there more behind it than that 

Jepson Taylor: The, it, it’s very specific. So what they did, the big breakthrough, cuz the thing you’re hitting on is they had the GPT models before GPT one. Mm-hmm. , GPT two, they were in the news and there was, there was some excitement about these models that, oh, these models might be dangerous because you can’t tell if a human’s writing this content or if AI’s writing this content. And the big breakthrough came when they fine tuned a question, answer data set. So now this becomes a knowledge retrieval opportunity. cause so you having AI write some Shakespeare or write a pretend post in the New York Times, that’s really not that helpful to either of us. But if you have it doing knowledge retrieval, so anytime I say a word that comes across as jargon or you’re like, oh, I don’t really know if I understand that, you can now go to JT PT and say, explain this concept to me like I’m five. 

Jepson Taylor: Mm-hmm. Now explain it to me like I’m, you know, 20 now explain it to me like I’m a PhD postdoc. And that is game changing. Like, think of all the books that you and I had to read that were boring. Like for your, for your degree. Like thermodynamics, physical chemistry, electrodynamics, these books were terrible, terrible books. Mm-hmm. mm-hmm. . They’re not written by expert storytellers. But now with the future generation, everything can be written by, do you want Dr. Seuss to read you electrodynamics Like mm-hmm. , that’s all on the table. And so it’s, but mainly it’s the knowledge retrieval. The analogy I give is, you and I went to school when we would do research, we’d have to go into the library and go find a few books and Google came around, which allowed you to do that a little bit better. But this is different. This is it. Read the entire library. Yeah. And it can produce exactly what you need. Yes, it can hallucinate, but as long as you are validating what it’s saying, it’s game changing. 

Dave Mariani: You know, when you, so, so obviously it has implications for search, you know, what do you think the implications are for search Do you think it’s, it’s an additional tool in the toolbox Or do you think it changes the game altogether 

Jepson Taylor: I, I think it changes the game so that, it’s so funny. If you, if you and I were talking a year ago and you said, how do you like Google I’d say I love Google, it’s great. I use it all the time. Mm-hmm. , now it feels like, I’m going back to 2001. 

Dave Mariani: Mm-hmm. 

Jepson Taylor: . So once you’ve been spoiled by, oh, I have this coding, like for developers, it’s the most, painful contrast. Cuz before if I’m writing code and I get an exception, what do I do with that exception Well, I either address it directly if I’ve recognized it, but if I don’t recognize it, if I don’t know what to do, I will copy the exception. I’ll paste it into Google and hopefully I’ll find an entry and stack overflow and I’ll read between the tea leaves and 30 minutes or an hour later, maybe I’ll have progress. GPT four immediately recognizes the exception and fixes it for me. It actually writes the, the code that fixes it more than it doesn’t. So it’s, I I see it as being a complete game changer. it’ll be fascinating to see if Google can catch up to open ai. I know they’re definitely trying. 

Dave Mariani: Yeah. Well even maybe the, the business model will change, right you know, when you think about even it’s already changed even for product search. You know, I, I use, if I, I got, I got my Amazon account, I’m just gonna buy everything from Amazon, so why wouldn’t I just search it using Amazon So that’s that, that was already changing. But you know, when you, when you look at it just a few years ago, you would say, who would be able to break through the breakthrough You know, Google’s business and owning search. Well, you break through it by changing, changing the game. Just like the internet changed, the game for Microsoft Windows became, you know, irrelevant in, in inter internet fa in the internet wave. It’s like, it’s, it’s seems like the same thing is happening, but even more so with G B T, where everybody is using it and everybody is afraid of it. So let’s talk about the being afraid of it. What’s your, what’s your opinion, Ben, in terms of should we be afraid of, of, of its capabilities and, and what it could do both good and bad 

Jepson Taylor: So before I lean into that, there is, you can think of AI as being a catalyst. So it allows you to do things faster and that’s good, but it can also be bad. It can allow you to do, subpar work faster. So there was that example in the news of that lawyer actually bringing some claim, some referenceable claims into the courtroom that were halluc they were made up. shame on them for not fact checking what it was producing, but just like you can use these types of technologies to do, to check your grammar, you can also use them to check for exploits opportunities. And so there’s definitely a major concern around leveraging these technologies to find vulnerabilities faster in open source libraries. And one of the things that’s missing today that I think they need to implement be these technologies are so powerful, you need accountability. 

Jepson Taylor: And so I, I would argue that the prompts that you are putting in to OpenAI or that I am, they should be you, you need to have better user verification. Cuz right now I can create a fake email, fake phone number, and I can go do things until I get shut off. And that should not be allowed. It, I, I should, there, there should be a level of discipline where if I do something illegal, writing some malicious code or doing something that has negative consequence, my open AI or my AI logs should be, able to be audited and brought into a court to prove that, hey, we, we know it was you. But right now that’s, that accountability isn’t really there. 

Dave Mariani: Data privacy is 

Jepson Taylor: A concern too. Sorry. 

Dave Mariani: You know, verified identity would help in so many different things, including social media, right I mean it’s, it’s, it’s, it’s harder to be a bully when, you know, you, you could be called out as such, you know So I, I don’t understand why we don’t have verified identity on the internet. why it’s so easy to to go anonymous. 

Jepson Taylor: Yeah. Well it, it’s interesting you, so there, there’s actually been a ca there was a case this year with bullying on Reddit mm-hmm. and someone on LinkedIn was able to look at a screenshot cuz this was, they were bowling LinkedIn people like, you know, look at this person on LinkedIn. And it’s funny, I, I like show up there a few times. Like, look at this, look at this dipshit, look at what he is saying. But this, someone was able to look at a screenshot and someone was saying terrible thing, terrible things about some of the women on LinkedIn. And so someone was able to figure out from a screenshot that that’s a real person. And this individual, they had to delete an account because when you were able to find, and you realized Eric had been saying online and you can’t deny it, it yeah. So it, it’s interesting how, how people change if there’s a mask. 

Dave Mariani: Yeah, it does. It definitely, it definitely, you know, politeness goes out the window when there’s a, a mask and a lot of other things. So I don’t, I don’t know. It’s, it’s if, if, if, if this, if this advancement sort of brings that to bear, that would be a net positive in my, in my view of the, of things. 

Jepson Taylor: Yeah. Well there, there’s a lot of net positives too on the horizon. Like the, the transformations that we’re gonna have in healthcare there, there’s also a lot of emerging threats that we need these technologies to help us with. So think of like CRISPR gene modification, like there mm-hmm. or even the, the risk of a singularity being the, the super intelligence. We, there are arguments that we need these technologies to try to help us get ahead of it. 

Dave Mariani: Yeah. I, I tend to always, whenever there’s a big leap in technology, I always, you know, I always say, you know what, it’s, it’s always progress and, you know, we’ll the, the things that we automate a way we’ll find better things to do. I’m, so, I I I hope this is, I hope this advancement sort of hues to that, that that logic, I think it will. But, 

Jepson Taylor: I think we’re already seeing that because thanks to technologies like Mid Journey, everyone’s an artist. Everyone’s a designer with G D four I, my 13 year old daughter, she was asking me how she can make more money and by more money, it’s not like house chores. Like she wants to make like five grand a month, which I think is hilarious, . It’s always telling her, well, you could learn to program. And she says, I don’t know how to program. And I started laughing. I said, yes, you do. And so I spoke into my phone with the GPT four app, write me my first Python script, and I was showing it to her. And so I said, like, look like everyone’s a programmer. Everyone’s, a journalist, like, not, not a, and I’m not saying this, that this will replace those jobs, but it opens up a new level of creativity for everyone. People can do more creative things. 

Dave Mariani: Well, you just said it, you said it earlier on that, you know, you spent time creating tools to automate the mundane aspects of your job. And I, I, I think this is, this, this is similar. 

Jepson Taylor: Yeah. Well, there, just like we have a limited number of heartbeats in this life, I like to think that we have a limited number of useful cognitive cycles. And the more you can spend, like I, if you’re working in a job where your brain is operating at 10%, it’s half awake. Cuz you’re just executing a process in front of you. I think that’s unfortunate. I would, I would hope that we would fight for more people to, to be challenged to their limit. because if you’re challenged to your limit, it means you’re gonna have to read a book, read something mm-hmm. or you’re gonna have to go across the hall and ask your neighbor, Hey, can I, I need some perspective on this and that. And, and for me, those have been the most exciting parts of my career. The parts I wasn’t qualified to do, I needed some help. 

Dave Mariani: Oh yeah. Hey, learning is always more, is always more exciting. So it’s like as when I’m not learning, I’m not happy. So, you know, that’s, that, that, that’s, that’s my attitude. But so let, let’s, let’s pivot to talking about what this means for business. So is business ready for this, this, this new sort of capability Ben, 

Jepson Taylor: It, it doesn’t matter if they are, cuz it’s happening whether they want to be or not. And it’s happening faster than we’re all ready for. So the, the transformation that is coming that we’re already seeing, it’s gonna be throughout the entire organization. So , it’s not just gonna show up in lead gen, it’s gonna be every aspect of business, product, marketing, sales, customer success, executive functions, every aspect of business will be impacted. And, and that’s because of the usefulness that we’re seeing. So if I wanted to make this hit everyone, I think you’re gonna see tools that come online that begin, summarizing emails. Mm-hmm. . So like there’s some executives that get 500 emails a day, that’s a nightmare, even for an executive assistant, that’s a nightmare. But having AI that can summarize everything and have auto-generated responses that are actually in your voice. 

Jepson Taylor: Now I’m not talking about the garbage that you see today. Like real authentic, like real responses that feel like Dave is writing it. and then you just go through these quick approvals or having your AE approve some of these responses mm-hmm. , that’s game changing for everyone. I am, this might be showing too much of the human side of me, but I am very excited for some of the impact disruption that will happen on the legal side. So when I sold my company, it was a thousand dollars per page for legal review, 90 page deal doc, I don’t think I’ll ever spend that again. Like mm-hmm, because you can have these technologies summarize, explain legalese to you like your five, and then have you write legalese back. That doesn’t mean that I’m not gonna use a lawyer on my contracts, but it definitely means I’m not gonna be spending 20 hours worth of legal. I’m gonna spend one. And I’m very happy about that. That makes me happy. 

Dave Mariani: You know what’s like, what’s interesting is that the definition of data has really changed, hasn’t it Because, you know, the data of course models need data. And again, back to sort of the, the old way of doing AI data typically was mostly numbers. I mean mostly structured. I know we always say it’s un Oh, unstructured data. Oh, unstructured data, but still mostly structured, right That went into training those models. Now data is everything, isn’t it It’s like, it’s anything, it’s anything that you can find a pattern with, which obviously means text, but that’s really like, the whole, you know, the whole internet is data. it’s not just data that’s stored in databases that makes us all work. 

Jepson Taylor: Well, data’s even more profound now than it used to be. So data in 2023 is everyone’s competitive moat. So if, if you’re, if you and I are competitors and if I decide I’m gonna come after you and hire a bunch of capital and go hire 10 people from Stanford, what is stopping me from doing that For most companies it’s going to be your data, cuz if, if you have more data than I do, but ultimately it comes down to whether or not you’re using it. So if you have massive amounts of data siloed away and you’re not using it to make better decisions, then you have no competitive moat. But if you are using it, then that’s an impossible moat for me cuz where am I gonna get the data that you’ve already accrued So I think d data represents a business’s experience and it is the competitive moat. And I think 2023 is the year that we all wake up and realize the power of data for, for every business. Well, you, 

Dave Mariani: Well you just said, you know, in your example of, reading, basically reading my emails and coming up with intelligent responses, I mean, I don’t think people really think of their email, their inbox as data, but it absolutely is, right It is training data, it’s training to be able to training, training the, the AI to be able to make those responses in an intelligent way. so, so that’s, that’s a really a game changer and a real, a real, a real mind twister. something that you’re just, you know, waking me up to in this, in this conversation here. you know, what is, you know, so what is it that, you know, you, you have commercial customers working for dataiku, and as a strategist, what are the kinds of things, Ben, that you’re telling, you know, your corporate customers about how they should be getting ready for this or what they should be doing differently now that this is upon us, 

Jepson Taylor: So for the customers and prospects that we interact with, that a lot of them are thinking more at the C level about AI transformation. And one of the fun things that we introduce into that conversation is the transformation side has been happening for quite a while. So we, we have customers like GE Aviation, they have thousands of models deployed. They’re, the opportunity was all always there, but now the attention is crossing as well. And so mm-hmm. . So it’s exciting to interact with companies that are really looking at culture changes internally. Cuz ultimately to get to an AI transformation success story, you want a culture change. So you need a top-down mandate coming from an executive and you need bottom-up evangelism and you need to have process that allows for innovation for processes to evolve. But ultimately you also need attribution. You need roll up, you need accountability. So there are a lot of data science teams that do a lot of activity, but they don’t have a rollup, they don’t have an attribution, they don’t have a defensible story to the cfo and then they just become a cost center. And so mm-hmm. , it really is the, the total picture, the total maturity that we see and then educating the markets on that. so, you 

Dave Mariani: Know, where do people like, so where does, where do you start I guess you start, you said you start at the top, that’s the location for where you start. So what do you, what is, what’s it’s, it’s cause it seems like a, it seems like a pretty complex, equation to solve for. So where do you start 

Jepson Taylor: Yeah, I love that you asked that because they’re, because of the usefulness of ai, especially today, if we went into a bigger company and did a full audit, there are over a thousand things they could do, like several thousand things they could do mm-hmm. . And that might sound like a good thing. It’s a bad thing and it’s a bad thing because they might waste time on projects they shouldn’t have started on. And so I like to think of, you wanna prioritize the projects based on feasibility. So feasibility, what’s the data accessibility Is it, is the data ready to go Do we have a process owner attached to it or domain expert, or is it gonna require some type of data migration first So data feasibility and then value, we wanna have some value assessment. Is this worth zero $1 sign, $2 sign or three, $3 sign would be something transformational. 

Jepson Taylor: The board’s gonna hear about it, it’s gonna impact your revenue by, you know, single digits. $1 sign for me is the threshold around a hundred thousand dollars. So if you’re working on a project and it’s worth less than that, let’s just give it a zero for now. It doesn’t mean you’re not gonna work on the project, but let’s not start on that one. And then the third one, I like the, the third one that came up was finding the right internal champions. So we have a project, it’s highly valuable, the data’s feasible, it’s accessible. Is the process owner a green vector, a yellow vector or a red vector A red vector, meaning they’re antagonist, they’re not supportive. Potentially they’re jobs at risk. And you wanna find projects that align on high value, high feasibility, but you’re also working with a green vector. Someone who they’re, they’re willing, they’re willing to go through this change process with you. 

Jepson Taylor: And if you can find a basket of those projects, three or five projects, that’s a great place to start. and those exist in most businesses, especially if the businesses are larger. and then over time, the the last point I’ll share with you is initially I strongly discouraged them from picking projects that could fail out of the gate. Cuz you want to get that momentum, you don’t wanna burn that political capital, but as you mature, I would then be critical of you if I came back a year or two later and if you had no projects that failed, then that, that’s an example of, an organization that is not friendly to innovation. So there, there is kind of that magic number, but I don’t want to have a magic number out of the gate. I, I want that. 

Dave Mariani: I love, I love that. 

Jepson Taylor: Yeah. 

Dave Mariani: Very, very cool. well I love your red vector green vector. so other than avoiding the red vectors, how do you turn the red vectors into green vectors when it comes to, people who are resistant to change and analytic leaders who could be, you know, blockers or antagonists really, to getting stuff done 

Jepson Taylor: I think that goes into the culture change, discussion. So one of the things that helps with culture change is to, to advertise some of the wins through the organization. There’s different ways to do that. If you did a monthly brown bag with data scientists, it would not be, it would not be well attended because data scientists tend to slip back into their deep jargon ways and mm-hmm. their data science posturing. But if you have a data scientist co-presenting with the business owner together on a quarterly or monthly basis, that’s gonna begin to produce some friendly competition internally, which I think is good cuz you can imagine mm-hmm. , if the marketing department is kicking ass or the sales department’s kicking ass, what does, what does that mean for the finance department Like they’re mm-hmm. , they’re gonna start to realize, oh, these leaders are getting, they’re getting multimillion dollar wins on the board. Why am I not And so that could be a good way to try to give some positive encouragement to these red vectors. but yeah, definitely avoid them on the projects you start on. 

Dave Mariani: I love that. I love that. Yeah. You gotta be a psychologist to work through culture change, don’t you As, and figure out how, what makes people ticks and what makes them, you know, what, what will motivate them. Yeah. so, so that, that, that, that’s some, some great advice, Ben. so, let’s, let’s, let’s, let’s talk a little bit about the, about the future and of our respective companies, right We got, you’re at Dataiku as a chief strategist, right in the heart and the center of ai. What are the kinds of things you’re seeing, Ben, that excite you, possibility wise or investment wise that you think, can really make a difference out there 

Jepson Taylor: so from data IQs perspective, they’re, if you, if you look at the different platform companies that are out there, you can kind of mm-hmm. put them on a spectrum from services to product and from seed stage to much more experience where you have a lot of use cases that you’re collecting. So Dataiku, they’re in this, they’re at a very interesting phase in the maturity curve. So they lean much more on product, which means they’re delivering on the democratization story. They actually have examples of customers using, customers loving the product. Something that was unusual when I joined at Iku is I saw a customer stand up in New York and give a toast with wine to the product. And I’m like, Hmm, I haven’t seen that before. But that’s good. That’s great. 

Dave Mariani: That’s good. That’s 

Jepson Taylor: Very good. Yeah. That means, that means the product is doing something. and then ultimately experience when it comes to applied ai, that is, that is the goal to be mined because it’s so many issues that happen with ai. Mm-hmm. models break production, some companies are successful, some companies are not. And so the more we can kind of capture all of that in this experience, net roll as much of it into product as you can, and then the rest into internal process, I think it’s a, it’s an exciting place to be. people ask who, who, who AI customers are, and it really is everyone. So focusing on the global 2000, it, it’s just like, it’d be hard for you and I to come up with a, a company that doesn’t use a database mm-hmm. , it’s equally, it’s, it’s nearly becoming as hard to come up with a, a company that won’t be using AI soon. 

Dave Mariani: So how does G P T sort of, affect the roadmap, for a company like data 

Jepson Taylor: So we Dataiku we haven’t really seen ourselves as a, an algorithm inventor. We’re, we’re really the kind of the, the backbone. We, we are an integrator. And so there’s really nothing about GT four that changes our trajectory. We, we have some, we have some announcements and press releases coming up soon. I’ll get my hand slapped if I say anything too soon. But in the next couple weeks, we we’re coming out with really good support for generative AI and for some of these use cases, and we’re premium inception partner for Nvidia, we’re taking that relationship very seriously. We were taking it seriously before G PT four, before chat G P T even showed up. And so I think Dataiku is the, the lack of technical debt that the platform has allows us to be very, quick to adapt to new, new technologies. So we’re, everyone’s geeking out about GPT four, but there’s always gonna be something new. There’s always gonna be some new tech stack, and I think Dataiku will continue to react and integrate quickly where it’s appropriate. How, how does it change, anything on Jepson Taylor What, what are you seeing on your end 

Dave Mariani: Well, you know, what I think is interesting is that, you know, our, the, the semantic model is the fundamental sort of, ingredient to getting customer value. because you can’t ask questions if there’s not a semantic model on top of your physical data. So that, that’s the first part of, of creating a, of a semantic layer is creating the semantic model. And I think that, you know, what we’re lacking, you know, what the semantic model lacks is, is code, is code to, is the data to actually feed the algorithm, to help train G B T to actually assist the modeler in creating that semantic model. So, I’m very excited that we have, we announced, a, a, a markup language, so we can address the analytics engineer to actually write code to create those semantic models versus using a gui. you could still do both. And so we can still address both personas, the one who likes to use a, an application versus somebody who wants to write code. But I think that the fact that we, now have, have code, we can actually use G B T to generate that code, to assist the, the modeler in actually creating the semantic model out of the box, or at least making, making it easier to create those semantic models. Yeah. So I think that there’s, there may be something there, Ben, but 

Jepson Taylor: Absolutely. 

Dave Mariani: Really looking into that 

Jepson Taylor: Well, maybe to geek out on that a little bit more, Dave, I feel like we’re all ending up in the same place. Mm-hmm. , so we’re all ending up where the, rather than clicking through a user interface, use natural language to request different things with visual feedback. So you can imagine a scenario where you’re at my home, we’re watching the sunset on the balcony, we’re drinking wine, and then I lean over my shoulder and I say, Hey, Jepson Taylor. And then like, , you know, this screen, I’ve got a TV on the balcony, it lights up like, Hey, what’s the status of this And I’m doing some things with my voice where five years before it would’ve been a massive project, like a month of work, even in Jepson Taylor. And so I, I think, I think that’ll be true with ai, with the semantic layer, it should just be just like you might task a, a senior or principal level engineer or product expert to go do something. You should be able to just ask the software layer. and then it’ll give you the visual feedback cuz you wanna make sure it comprehends, you don’t wanna Yeah. Go to bed and then wake up in the morning and yell at Jepson Taylor that . Like, that’s not what I wanted. You need some visual confirmation. 

Dave Mariani: Yeah. And I think we got a, I think we got the right ingredients to do that now. so I’m excited to see where we can go with that to answer your question. but, you know, I know it’s like it’s, I’ve taken way too much of your time, but I don’t wanna let you go without asking founder to founder. I mean, there’s a lot of, there’s a lot of future, entrepreneurs out there or entrepreneurs that are trying to do what they’re doing and be successful and you’ve, you know, you’ve seen a lot of that, Ben as a, as a founder. so what, what are some of the things you think, what’s some of the advice you can give to an entrepreneur out there some, some things that, you know, to avoid, some things to do. So some things that, so the dos and the don’t of being an entrepreneur, what would you say 

Jepson Taylor: I, that’s, that’s a complicated question cuz the, I I’ve been critical in the past of first time entrepreneurs cuz they’re just, like I I was that right So it’s so stupid. 

Dave Mariani: We all were that yeah, we all were 

Jepson Taylor: That and no, but we 

Dave Mariani: Knew we were like, mean, we look back and we say, how stupid were we when we the first time entrepreneurs So yeah, it’s not like we’re being critical to other people. There we we’re, we’re being, you know, 

Jepson Taylor: Well, I honest 

Dave Mariani: With ourselves, well I 

Jepson Taylor: Was, I was talking to, a VC and I was saying, yeah, these, these first time entrepreneurs, they’re just bright-eyed, bushy tail. They’re so naive and, and I, and I was kind of taking a negative tangent on them as like a persona. And this VC said they need that, they need that impossible reality distortion take over the world. They’ve got, you know, the next Facebook of their industry, they need that. And, and he had a fair point. It was like, well I guess they do cuz they’re quitting their jobs, like mm-hmm. , if, like, imagine if you’re a first time entrepreneur and you just quit your job and you meet me in a bar, and if I just take, like if I just like take the most negative perspective on if I could bet against you, I would, I’d probably make, you know, make my money back. 

Jepson Taylor: You’re gonna make these mistakes. You’re, you’re gonna have to pivot multiple times. if you have a co-founder, you, you know, you might get in a disagreement where you have to remove them. Like you, you could go down a list of like all of this crap that’s coming, oh, and you’re gonna have to fire someone in the next four months. And if you don’t, you’re an idiot. Mm-hmm. and like mm-hmm. , you could come up with a list of stuff where as a first time founder, I don’t know if I’d want them to be, I, I would just encourage them to have good mentors. Right Like, have good mentors. 

Dave Mariani: Yeah. You’re gonna make first time mistakes. I mean, it’s, the difference between experience and not experience is like you’re gonna, you’re, you’re gonna step into some potholes and once you’ve done that a couple times, you’re, you’re smart enough to avoid those on the next run. Yeah. 

Jepson Taylor: You did make, you did remind me of one that was especially painful. sometimes you’re kind of whip song between urgency and strategy. Like sometimes you’re con oh, totally consumed by urgency, just customer demands, anger, customer, you’re doing these different things. And one of the things I wish I could go back in time and kind of help myself out with is if the customer’s stream screaming for something that’s urgent. Like let’s say you’re, I was dealing with something that was completely consuming for a month. I wish I had enforced a moment of strategy just like, you know, e every two weeks, half a Friday, take it off with your co-founders. If money wasn’t an issue, if you could fire 20% of your customers, what would you do And it would just be a really good, I I think there would’ve been opportunity to just kind of have a fresh perspective. Cuz I think when, if you’re stuck in the a hundred hour work weeks for four weeks just trying to appease a particular customer, you could be so off course on strategy, but also you’re, you’re, you’re not lifting your, you’re you’re not lifting your view and looking at the horizon. You’re just stuck chasing your toes. You’re, 

Dave Mariani: You’re reacting, right You’re reacting. Yeah. As you know, as opposed to, you know, being proactive. You’re just reacting. and you’re reactive and you can’t, cuz you don’t have the time to do anything otherwise. yeah. So like, my, my little trick Ben, don’t tell anybody, but my little trick is I block out time on Fridays in the afternoons, I just put a big, a big block on my Google calendar and people still try to violate that and, and I’m okay with that, but at least it gives me some white space. and I save that time for doing new projects, for writing code or to, you know, figure out something, that is on my list to do, to figure it out. but without having, cuz you’re right, you know, being an entrepreneur, being the I was the CEO and having to worry about every aspect of the business, you, it just, it’s just impossible to do anything really, really well. 

Dave Mariani: you gotta have and trust, you gotta hire the team and trust the team to do it because all you can do is just be unblock things for them at that point. but, but without the ability to really think things through, you know, you’re likely to give your customers what they want. and what your customers want is not necessarily what they should want. because they may ask for something that is a fix to a f to an architectural flaw, which is just gonna make, you know, increase your, you know, your tech debt, versus coming up with a whole new approach that will fix not just that problem, but a raft of other problems. So, and you can’t really get that unless you take a step back and you have to have time to take a step back to look at it. So that’s really good advice. 

Jepson Taylor: Yeah. Well it’s, I I I love the startup journey and I, I I hope more people attempt it. One of the things I tell people is when you go to do a startup, you need to chase an idea big enough that you’re willing to fail. And then I define failure. So failure is foreclose into your home, bankrupt max out credit cards like a have to ask friends and family for favors. And when you say, when you say that to college kids, you kind of get, you see this reaction on their faces, like, oh, I don’t know. Like, you see the hesitation and then I go back to big enough and I say, and that’s why it’s so important that the idea is big enough for who, not for me, not for your professor, not for your parents. Big enough for you. So if you find, and I, I think I noticed that I’ve always wanted to do a startup and it wasn’t until I found an idea that was so big, it forces you to do something irrational. And, and that’s, that’s exciting. I I would celebrate that for anyone listening. Find that idea. And it’s not the end of the world if your startup fails. It’s, it really isn’t. 

Dave Mariani: And you know what, and to get to find that idea, go into business doing something and then, and keep your eyes open and look for gaps. I mean, for me that’s how scale came about. Is it, it wasn’t in the shower that popped into my head. It was, you know, years of slogging through trying to make data, you know, consumable for people. but you know, one other thing you mentioned, Ben is like, is this about when you’re young and you’re a first timer, you don’t have a mortgage, right You have rent. Yeah. you don’t have kids yet typically, so you don’t have to worry about supporting kids and putting ’em through school. So it does allow you to swing from for the fences much more than, when you do have a mortgage, you do have kids and you do have college where the price of failure becomes a lot more acute. And it’s not just your failure, it could be your whole family’s failure. 

Jepson Taylor: Yeah. So, well, I, I’m gonna give your listeners some bad advice. So , I, I completely agree with you. If you’re a a young college kid eating ramen, no better time than to go to a startup cuz you can actually, you know, 30,000 a year might be a pay raise for you, but if you’re older you can actually burn through your 401k and do the same. So that’s what I did. maybe not advise, but you, you could still do the same, just, just dip into savings and, and running. 

Dave Mariani: Yeah. Yep, yep. Well that’s, yeah, that’s, you just, you just called out me right there buddy. So , thanks for the, that’s what I’m doing. . 

Dave Mariani: It’s all gonna work itself out, right Ben So, it’s, yeah, it’s, look, it’s, and it’s amazing and it’s so much fun and you have to love what you do. cuz you’ll do it all the time. and, and if you do it all the time, you’ll be good at it and eventually you’ll figure out how to make money. That’s, that’s, that’s my advice to, to the listeners out there. Ben, you have been awesome. I always can talk to you for hours. we can go in so many different areas. so everybody out there, Jepson Taylor, follow, follow, Jepson, follow Ben. he writes some great stuff, and puts out great content that you can learn from. So, so go and, follow him on LinkedIn and follow him wherever he goes. So Ben Jepson. 

Jepson Taylor: Yeah, thanks Dave. It’s been a blast. Always, 

Dave Mariani: Always a fun time. Yeah, 

Jepson Taylor: Absolutely. 

Dave Mariani: Take care and everybody out there, stay data driven. Have a great day. Absolutely. 

Be Data-Driven At Scale