Vin Vashishta explains why we should stop using dashboards
October 4, 2023

Vin Vashishta explains why we should stop using dashboards

No items found.

Vin Vashishta, the guy we all love to follow, has never seen a dashboard with positive ROI. This time on The Data Engineering Show, he met the bros to talk about the difference between BI dashboards and analytics that actually introduce knowledge. It’s no longer just about the data volume, it’s about quality and relevance.

Listen on Spotify or Apple Podcasts

Benjamin: Hi everyone and welcome back for another episode of the Data Engineering Show. So today we have Vin Vasishta joining us, which is awesome. Good to have you on the podcast, Vin.

Eldad: Hey.

Vin Vashishta: Thanks for having me. I appreciate it.

Benjamin: Awesome. So for everyone who hasn't heard of Vin and Vin is really well known in the kind of data thought leader space, he's an AI advisor, kind of co-founded V Squared, which is a data and AI consultancy and wrote a really well known book called From Data to Profit. So great to have you on today. We're super excited to chat and joining me of course is the data bro, Eldad. Good to have you back after being away for the last two episodes. So welcome back on the show.

Eldad: Thank you for giving me another shot on the show.

Benjamin: Yeah, we discussed really like a long time before what we said. Okay, we'll, we'll have you back. Awesome. Uh, cool. Vin, do you just want to quickly introduce yourself, uh, kind of, uh, yeah. Uh, and tell us what you're up to these days.

Eldad: Yes.

Vin Vashishta: Yeah, definitely. So you covered the high points. Vin Vashishta, been in the technology space for almost 30 years in data science and machine learning for the last 11 and a half years. Went to school to do data science and graduated in the 90s when no one wanted it. So I had to go into software development, software engineering roles, built and led teams, got really close to implementation and execution. We had to deliver more than most data science teams do.

because the technology was more mature. So there's a closer connection in my background to implementation, to execution. I brought that with me in starting vSquared. Really quickly realized that being successful with digital and with software and with cloud and all of the other technology trends, very different than being successful with data and AI. So I built my consulting practice around not only implementing solutions, but all of the other components that are necessary.

Really the book is the culmination of 11 years journey through how do you make this, this technology actually result in some cashflow for people. So that's, uh, that's my background. What am I up to today? A little bit too much. I'm not going to go into everything. I'm doing too many things. I'm going to put it that way. I need to take a vacation.

Benjamin: Nice. So you said that you.

Eldad: Data is hard, data is not easy. It's not like the 90s.

Vin Vashishta: Not everyone can do it. No, I think data was harder in the nineties because we didn't have any of the infrastructure and none of the libraries. I mean, we were trying to build models and see it was, it was painful. Yes.

Eldad: It was all about being creative in the nineties, like the movies in the nineties, like, you didn't have a budget, you had to be creative and then, you know, make assumptions about the data and every now it's all about facts and it's boring. And, but yeah, I mean, that's good.

Vin Vashishta: Ha ha ha ha ha. Yeah, no more shooting from the hip anymore. You have to actually be able to back up what you can say you can deliver.

Benjamin: So one thing I'm curious about is you said you like over the time at V squared and also before that realized that running successful data projects is kind of very different from running successful software projects from a more traditional software background. Like tell us more about that, right? How is it different? Like what are, what are your lessons there?

Vin Vashishta: Well, I think the biggest difference is the way that they're monetized. So the companies that I worked with in the early 2010s, they were solving the technology problem. And the first couple of years that I'd started vsquared, I got swept up into the same thing. I was helping them solve technology problems. We would deliver really successful initiatives, but trying to maintain momentum, trying to get that, you know, the hype, the, the tens of millions, hundreds of millions, billions.

It was more complex because we needed bigger budgets and we had to then justify. We had to support it better. And what you realize very quickly is you have last mile problems, which are the technology problems, but you also have first mile problems, which connect to the business strategy data is unique. It has unique monetization properties. We have to gather it in different ways. We've been gathering it, really thinking about business intelligence use cases.

And to use it for analytics, for machine learning use cases, completely different data gathering, data generation, data engineering. So all of those are really enterprise wide problems. When I broke it down into pieces, realized you have to fix problems that start at the strategy level, move into culture and then go into technology. And data teams inherit. I don't know.

10 to 20 years of technical debt from every other organization and every other part of the business. And we're expected to solve all of these problems. Some of them are technical and we can solve them, but other ones are cultural and strategic. So we need new roles. That's what I began filling myself and now I'm teaching people how to fill those roles and I teach companies how to build teams in a way that really connects all of the dots, not just the technical dots.

So very different monetization, very different development. If you try doing it the digital way, it's a failure. It just won't work.

Benjamin: And so just for me to get this, when you're talking about monetizing data products in this case, is this always something customer facing, right? Say a new ML recommendation system for products in your online shop or something, or is it also for internal use cases?

Vin Vashishta: I think we need to treat all data and AI products like they're customer facing, even if they are internal facing, but it's both.

Benjamin: So I get it for the shopping card recommender. You can say, hey, we made this much money and kind of feel like this user group, I do an A-B test, I see, hey, 10% profit, easy, kind of, right? You can show the numbers, say, hey, we did amazing here. How is that for internal data projects? Because there it seems much harder, right? I'm kind of working on some dashboards for someone else to use. Like how can you approach monetization? They are just kind of putting some...

Business value onto that.

Vin Vashishta: When you look at dashboards, it's a good case for self-service data tools, because I've never seen a dashboard that had positive ROI. Your data team, the cost, yep, whoops, whoops.

Eldad: What? Wait, we need to freeze, we need to pause for a moment, yeah.

Benjamin: What? Eldar, Sisense was useless.

Eldad: We never cut anything out of our data pro blog post, but you're so right. You're so right. Dashboards are just destined to die eventually. And then, but they stick, they stick around. I've seen dashboards that were running for years, years and exactly for the reasons you've mentioned, culture, alignment, you know, having three Salesforce accounts with having three teams updating those three different accounts. And then you need a... to make sense of a data lake. And it's hard, it's hard. And you're right, it's about culture and we all wanna be data-driven, but it's not enough anymore to be just data-driven. Yeah.

Vin Vashishta: Yeah, I think the important part about the dashboard, especially that component of it, we should give users the power to do some of these things themselves. And we underestimate users, consistently underestimate them. I have trained users who have never done any sort of engineering before. I've trained people in marketing teams, especially, to build out dashboards using Python, Jupyter

connecting to SQL, it is so simple now. And we underestimate what users can do on their own. Low code, no code tools are good, but we can actually, when we talk about data literacy and AI literacy, we can train frontline users to do so much more. And we can then support them in a way that's feasible. Behind dashboards, that's where we can deliver some ROI. It is...

Sometimes those models powering the dashboard that you never see, never think about, never hear, those are where the true insights come from. When you look at what it is that data scientists, data engineers, data analysts should be doing, it's the more complex. That's what we take on. If you want someone who simply does a select statement and displays the output, there's no way to make that a positive ROI initiative.

And so data, you know, frontline users should have the power to be able to do that for themselves when it comes to. Oh yes. Yes. Let them have it.

Eldad: But they're so addicted to dashboards. They're so addicted. Because we taught them, everyone taught them consensus on data is more important than anything else. So everyone is waiting for a dashboard because of that consensus that needs to happen, right? The single version of the truth, like all of those philosophies. And it's pretty depressing if you're on the go-to-market team and you need to support your customers.

Vin Vashishta: Mm-hmm.

Eldad: And you need the data to do it and you get a dashboard and you ask, like, can I drill into the numbers? Like I see that average, right? Like nobody wins on averages and dashboards are all about showing average numbers. So people want to drill in, people are looking for a way to express themselves with data. And that's a big question because there are so many religions out there. You've mentioned Python, like the languages, right? How do we approach data? Should I like, and that's changing. And technology gurus tell us like, oh, we just had Scala thrown out the window, like just, right? Like it feels like 20 years, it happens a few. So it's SQL, the way to do it is, is there, is AI going to help us kind of figure that out, how we move to natural language? Cause language is barrier, right? Like interfaces is barrier for users and cost obviously, but cost is being tackled by technology, so we should assume cost will be lower. But.

Vin Vashishta: No. Ha ha.

Benjamin: Thank you.

Eldad: interfaces don't really change. And that's the dashboard. We were stuck for 20 years with the same pie chart. And then I love pie charts.

Vin Vashishta: Well, when you think about dashboards, that's the BI mentality. That's the digital data mentality. You gather data to display it. And so it's gathered close to the people who will need it the most. And it never escapes that silo. So dashboards aren't data science. Dashboards aren't analytics. That's BI. And we have to stop saying dashboards are data science. Dashboards are analytics. They're not. They really aren't. I mean, an average.

Eldad: What makes BI from data from your experience, if you look at the data in the metadata only? You don't even know who the users are. What has changed? It's the same builders, the same builders that build the BI stacks before. Now they're moving forward to do something else, something bigger. Taking over engineering, owning the business is owning the data.

Vin Vashishta: Mm-hmm.

Eldad: So kind of from your experience talking to so many companies over the last few years, what's changing there in terms of data politics and like...

Vin Vashishta: Data politics is a thing now, but when you look at, so the challenge here is, and I don't wanna make anyone else mad, I've already probably made about 90% of your audience mad, and I think I'm gonna go get the other 10% right now. The people who created this problem are now trying to sell you the solution to it. The people who created all of these digital data silos with BI tools are now realizing that analytics and data science, yeah, hoo.

Eldad: Who's that? Who did that?

Vin Vashishta: No, not micros. I mean, no, but looking at, you know, all of these companies that created a digital data infrastructure. Where if I'm sales, I gather sales data and I keep sales data in the sales database connected to a sales app or an ecosystem of sales apps, because why? Why the heck would HR need any of that? Why would anyone else need it? It's sales data. It's us. And there was very limited cross pollination. That's BI. That's digital.

When you come into data for analytics, for machine learning, now we're looking at longer chains. Now we are looking at, instead of gathering data, we are gathering domain knowledge and building out a domain graph or building out a knowledge graph is the best way to manage for our use cases because that's the value. Why go from BI and dashboards to...

analytics and machine learning models. Well, what's the ROI? And that's really the, what's the difference? Why should I spend more for an analyst than for somebody who's doing BI? Why should I buy this new application? And I think it's important for companies to ask those questions because those are the only ways you start the discussions that lead you to this is different. This is going to find patterns that aren't obvious. This is going to introduce domain knowledge into workflows that we haven't had before.

Sometimes that means we're going to be able to make better decisions. Sometimes that means we're going to be able to see forward further. Sometimes that means we're going to be able to take in all of these complex symptoms and diagnose a root cause and understand the implications of potential fixes. That's not something BI handles. And in order to make the move, we have to stop thinking about it in silo monolith BI dashboard.

and begin to think about holistics, systems, knowledge graphs, domain knowledge that's new, not that's existing, that's new, being extracted from the data using patterns that aren't obvious. They're there, but it's so much complexity, so much trash, so much, you know, all of this other stuff that it's not obvious to people. So we use the math to tease that domain knowledge out. And introduce it back to the business or introduce it into automation so it can handle different parts of workflows for business users, for customers. So that's different and we have to justify it in order to start the conversation. And if business leaders start asking that question, why? Why can't I just use BI? Why can't I just code this up? Why do I have to go to this next level? Then we have those conversations. You're forcing us to start with value.

Benjamin: So in that world, like how, how does a data team actually operate? Right. Then where do these types of projects then come from? Right. Now I have a business leader coming saying, Hey, I need a dashboard for, for XYZ in this kind of a different world you're describing how, what types of questions kind of who asked these questions.

Vin Vashishta: Yep. The biggest problem is if your boss asks for a dashboard, just say no. And aggressive staring. No.

Eldad: Or more important, how do I answer my boss that they don't need a dashboard?

Benjamin: Hehehe

Eldad: Let's quit.

Vin Vashishta: So the best way to answer that question is to explain none of this sounds like data science, does it? None of this sounds like data engineering, does it? Doesn't sound like analytics either. So let's stop forcing data scientists to do all this stuff. This is one of the primary problems. We need data literate business users. We need AI literate business users. We also need business literate data scientists and data engineers and analysts. So that's important. But we can't, and we're doing this in both sides of the equation, we can't overstep this and start asking all of our business users to become data scientists and analysts. We can't ask all of our data scientists to become product managers and strategists.

We need new roles. We need new roles in frontline organizations that are non-technical who are these hybrids where they are domain experts and they have the ability to build these dashboards. They have coding capabilities. They understand SQL. They understand hopefully one of the no SQL databases too. Let's be a little modern here. And you know, that would, that's.

Eldad: So you're a company and you spent the last year searching for data scientists because like, you know, that's what they said last year and there's no one to find, right? Like your whole business is based on cold calls. And right. And then that's how you feel, at least as a unit manager and somewhere, even at the most techie company. And how do you transition teams? How is the future going to look like? How do we going to like, what should we ask for when we recruit?

Vin Vashishta: We have to pick up these new roles. You can't expect to adopt an entirely different paradigm of product without product managers. You can't implement technical strategy, an entirely new type of strategy, without technical strategists. We can't expect CXOs to magically become these things that they've never been before, especially when the role is as big as these are. So we need new roles. Like I said, in the frontline teams, we need people who are hybrid.

Benjamin: Thanks for watching!

Vin Vashishta: They are domain experts with technical capabilities. We need in data teams and data organizations, we need people who are product managers and strategists. We need a top layer, a C level layer of leadership for the data team, who is not just a people leader, but also a leader of strategy, where they can take the AI strategy and implement it. We have to accept that this is not just a technology problem, but it's also not enough

To say technology team, all you do is technology. They also need to own product. They also need to own strategy, whether that lives in the data team or in a product organization or in a strategy organization. That's, you know, organizational structure is really what works best for the business. Big businesses, small businesses structured differently, different industries, there are different structures that you're going to put together. But.

Benjamin: Thank you.

Vin Vashishta: You have to acknowledge there are new roles that are necessary to succeed. Isn't enough to just throw technologists at this problem. It won't solve the value side of the equation. It just gets you a whole bunch of technology.

Eldad: Because you're not throwing enough technology at the problem. Just throwing a... Just enough technology will not solve it.

Benjamin: So.

Vin Vashishta: Let's get some AI into the product manager space and into the strategist. Let's hire GPT as our product strategist and see how that works. You and me, we got a patent. We got to get together after this. We can get some funding for that.

Eldad: It will be a big thing, it will be a big thing, I'm telling you.

Benjamin: Nice. So like, say I'm a listener, I'm listening to this podcast and saying, Hey, this sounds cool. And I'm giving you a call, right? I'm saying, Hey, like Vin kind of V squared, like come to us, implement these things. Like, how do you actually approach this? Right? Cause like what you're describing at a high level seems to make sense, but actually getting this into an organization, like there's going to be a lot of friction. Right? You need kind of buy in from the highest levels. You're kind of talking about big changes. How is that actually, I guess that actually working?

Vin Vashishta: Mm-hmm.

Eldad: Nah.

Eldad: So Vin is coming in after the changes and then everyone is ready to listen. I think like, right? No, I can need to know when to... Yeah, go ahead.

Vin Vashishta: Yeah, exactly. Yeah. If you are in a data organization and you're thinking about bringing me in, you're at the wrong place. It won't succeed. You really need a strategist, need a product manager to begin the process with your C-level leaders. You need buy-in from the top level. And that can come from two different directions. One, you can.

establish a track record of success, just deliver some, you know, they, they call them quick wins, but they're not so quick. And the win is much bigger than most people expect it to be. Deliver a couple of those that actually have top and bottom line impacts. You get attention from CXOs. They, they will show up because growth isn't easy anymore and they're on the hook to save costs. So if you're doing both, they will come to you and that's when you can make the pitch and say, we need more.

If you really want the top level value, you've seen nothing so far. We can do a lot more, but we need new roles and we need people to come in who can help establish this connection across the enterprise. The other way you can do it is by scaring the daylights out of your CXOs, because they already are scared. If your C level or founder has made some, some wild claims about how they're using AI and going to

drive growth with it. Um, we all know many of those claims are not materializing as fast as they've promised and they're, they're running out of time. Investors are punishing companies, especially if you're a startup. Oh yeah. If you're a startup or if you have, uh, you know, if you're a multi-billion dollar company, if you've said the AI, yeah, if you've said the AI story, but you have not delivered the AI results yet, and that means, you know, money, cash.

Eldad: Listen to that Benjamin.

Vin Vashishta: It has to show up. If you haven't done it yet, your CEO right now is sweating. The board's calling. Investors are asking for more than a story. Yep.

Eldad: Everyone is all in on AI. Everyone is all in on AI. It's like has to succeed for everyone

Vin Vashishta: Yeah, they're all in, but now they want to see some cash. They want to see the chips show up. It's, you know, you can push all your chips into the middle of the table. If they just disappear, they're not going to keep coming. No one will continue to fund you if you keep throwing chips at nothing. And that's the, there's the duality. On the one hand, you can demonstrate how powerful the opportunity is. And on the other side, you can demonstrate almost like a lifeline or a life preserver.

Eldad: Thanks. Yes.

Vin Vashishta: where you reach out to the business and say, look, we've never had this relationship before, but we need to build it now. You need to bring me your problems, I'll solve them. Don't bring me a technology, don't tell me generative AI. I figured that out, remember, I'm the technology person. We will get to AI, but we're not going to get to it the way you thought we were. But we're still going to give you cash, and that's what your investors want. And that is the life vest I am going to give you. Would you like to take it? That's the second way in is a little bit of fear.

Eldad: Classic selling, you know, always fear the customer, always scare the customer.

Vin Vashishta: Mm-hmm. Well, I think if the opportunity doesn't work, you have to. I start with the carrot. Yeah, I start with the carrot and say this is the size of the opportunity. And if that doesn't work, then well, let me show you this scary monster in the closet.

Eldad: Yes. One way or another. Yes.

Eldad: Makes perfect sense.

Benjamin: So when, when you said have some quick wins in the beginning, get your CXO on board and so on, like that actually already assumes that you can show that your quick wins are driving profit in some way, right? Like that seems a bit cyclic. Uh, so how, like, how do I get to that in the first place? Cause then you're saying, okay, then you're

Vin Vashishta: Yes. Why would you start an initiative if you didn't know what it was going to return? I mean if I walked up to you on the street and said, give me five bucks, what would be your first question be? I mean, really, right? Yeah, that's, but I mean if you're on the street, I walk up to you, I'm just somebody that you've seen before a couple of times, and I say, hey, I need five bucks. You're not just going to shell it out. If I come over to you and say, hey, I need your help, come on, let's go do this thing. You're gonna say, what thing? Why am I doing this? Wait, hold on, slow down.

Benjamin: Hell no.

Eldad: Why just five?

Vin Vashishta: That's what we should do. We shouldn't just start working. We shouldn't take all of this, this stream of consciousness that's coming to us from the business and accepted at face value as being something that'll generate returns. We have sort of this magic aura around us right now because the business needs us. And it's a true need. So being able to push back is one of our superpowers. We can say, look, I want to return value to you. There's one of me, there's 80 of you.

let's figure out what we should be working on to get the highest returns. And if you all want something done tomorrow, let's hire some people. By the way, that's going to cost a lot. So you'd better have some ROI behind this. Let's start talking about this just in basic business terms. And if you don't have the ability to have that conversation, it's okay. You're a data engineer. You're an analyst, you're a data scientist. You don't have an MBA. That's okay. You weren't hired to have an MBA.

Talk about this in pragmatic terms and say, look, I don't do ROI, I don't do product strategy, I don't do AI strategy. Maybe we should get someone, maybe we should hire someone. And it doesn't have to be a consultant. I mean, I know this almost sounds like I'm pitching my services, but really I'm not. Train somebody into the role. There's tons of people in the business, in the data organization who want to go into these new roles, who want to own the product more.

who want to have more control over the direction of strategy, of that high level, where are we going to go with this technology, of developing a vision. There are people who want to do this, just upskill them. Don't pay me a really large number of dollars per hour to be in your business for over a year. Don't do that, hire some people. You will be happier in the end with that approach. And so that's the...

say this more often than I really pitch my own services, give your people career paths, they'll stay longer. Explain that we need new talent and you'll be more successful. Don't try to take on a role that you're not qualified for and half the time don't want to do in the first place. Don't feel pressured into that. Just start the conversation with value. And when the questions come up, don't be the engineer. We want to be the one with all the answers. Just say, look, I don't do this. It's like asking you to write something in you know, pick a programming language that no one ever uses anymore. If someone were to ask you to do that, yeah. Yeah. If somebody said, I need you to write an enterprise app and I don't know, Fortran and you'd probably, um, maybe we need someone else for this. I mean, I'll take a look, but

Eldad: VBA.

Benjamin: What's I never even heard of that? No, just kidding.

Eldad: on purpose, that's why I used it.

Vin Vashishta: Look at it the same way. If you're not a strategist, don't feel pushed into it.

Eldad: I'll take it to the bank.

Vin Vashishta: I'll actually get the five bucks. Can I have that five?

Eldad: I'll take that.

Benjamin: this. We, to maybe hit you with a more controversial question, we had Joe Rice on the last episode and he actually said, I'm tired of talking about profit all the time and ROI, right? And it's showing that as an, exactly that kind of as an industry, we're talking too much about this and actually shows that we're not delivering enough value and enough ROI because we need to keep cycling on the same thing over and over. Like what you, yeah.

Vin Vashishta: Mm-hmm. Yep. I watched that one. Yep. Yes. Yep. Mm-hmm.

Vin Vashishta: Right.

Benjamin: What would be your take on that?

Vin Vashishta: Yeah, he's right. I mean, it should be like chewing gum. Chewing gum doesn't need to advertise. It just tells you what flavor it is. Pepsi doesn't advertise. It's just a Pepsi. It is sitting right there on the shelf. It doesn't say here's my value proposition. It just advertises somebody drinking a Pepsi and looking happy. Like that's where I would want to get, but, you know, unlike Joe, I don't live in that world, Joe's kind of a rock star. And his book, let's just say we both have books, but his book appears to be doing a just...

just a little better than mine. So he's, yes, yes. But the people that I talk to don't understand it. And the reason why it's really two-sided, one side of the problem is they don't know how much it's going to cost them and how much work it is. It's enterprise wide. When you're looking at data engineering, just as a niche.

Eldad: Wait, wait! Patience, patience!

Vin Vashishta: You think it's just an engineering thing. If we move our data from all these other places to one place, everything's cool, but it isn't. And that's the, you know, if you read Joe's book, he's actually in that book. He gives you that bad news a few times. Yeah. Guess what? Most of the data you have worthless. And when you centralize it, it's still worthless. So you've just spent a whole bunch of money centralizing, you know, landfill. It's.

Eldad: to have one version of this centralized worthless truth. So at least it's one version versus.

Benjamin: It's the best landfill.

Vin Vashishta: But yes, I mean, it has to be truth, though. If there's trash in the data, it's not truth. You know, it's and centralized. Oh, yes. It's dangerous. Well, and centralization is wonderful, but we have to talk about centralization differently. We're centralizing knowledge, not data. The data isn't a display element.

Eldad: No, but that's true. The perception that data is oil is nice. It's a nice thing. Yeah.

Vin Vashishta: the data contains more complex domain knowledge. And we're centralizing it so that it is easier to take that domain knowledge out and begin to use it in multiple ways for customers and internally so that we can create and deliver about value more efficiently. That's the goal. And if we spend most of our time moving bad data around, it costs the same to move bad data around.

as it does to gather new good data. So why don't we just gather good new data? Well, it's because the business doesn't understand the value of it. It's expensive. And they think, but we already have data at home. And unfortunately, the data that they have at home is kind of like the dollar store knockoff version of whatever candy bar you wanted to buy in the store, where you wanted to buy the good bubble gum.

Eldad: You know, remember, remember swatch the watches. And remember when we used to collect them as if those will be like Rolex. You still die. Just some still do.

Vin Vashishta: Oh yes, yes. Used to. Wait, used to? I am the last member of Members Only. I just want to let everybody know.

Eldad: So yeah, so welcome to the club. And it turned out to be worthless, but it was a lot of fun. And the thing is people spend tons of energy on, go ahead. People spent tons of energy on building data pipelines. And now you're coming and you're telling them like, why did you do that in the first place? Like, why did you build all of those data pipelines that are designed to scale, designed to build lakes, designed to offload and unload and transform.

And that's strategy. So that was 10 years, all about that was the strategy, like getting the data in. Now people start to realize, okay, so that's worthless, or at least it's not as equal. As oil, so it's not more data, more value. It's just more resolution. And as you said, without the domain, it's pixels. Nobody can understand them. So.

How do you negotiate, given the fact that data is only becoming like managing data and utilizing data really becomes super complicated and users are going to work every day. They need to get the job done. So.

Vin Vashishta: Yep. Strategy has to be lightweight. Strategy should be a framework for decision-making that informs and improves decision-making across the enterprise about data and AI. That's what strategy should do. If you think that gathering data is a strategy, we're in trouble. And this is the problem. You know, when Joe says, I don't want to talk about value anymore. We want to get to the point where we understand the value of data gathering.

And that means we have to do that education step where we explain, you have to gather data differently. Gathering it for BI has BI value, but you didn't hire me to do BI. You hired me to do data science. Data science requires a different type of data. So if you force your data engineers into a digital paradigm and give them digitally gathered data, well, guess what you're going to get in that data repository. So if we start with tactics, if we start with technology.

All we get is more tactics and technology. If we start tactical, you're going to hire a ton of individual contributors. And a McKinsey study is, I think two months ago, where they looked at what companies who have several successful deployments in production versus companies who haven't been able to succeed with a deployment in production. The lower maturity, less production deployments.

valued AI researchers, data scientists, individual contributors the most when it came to roles. The people who had a ton, the companies who had several multiple production deployments, leaders, translators, strategists, it's a different kind of problem. So there's, we want to get to that point. And I think that was Joe's point is we want to get there.

to where you can just say, I'm going to build a model and everyone goes, yeah, I got it. Where it is so ubiquitous and we've delivered so much value. And he's also right about, we've been talking about delivering value for so long, we're in danger of hitting, you know, sort of crypto web three territory, where there's a whole lot of talk, there's a whole lot of hype, there's not a whole lot of delivery. There's not a whole lot of value creation.

And even if you're not a scam, you can sure get labeled as one.

Eldad: So you're saying everything that runs on the GPU is a scam? Everything you run? Any data-related project that ends on the GPU? We should be worried. But I'm not correlating anything. We love you, NVIDIA.

Vin Vashishta: I'm sorry, Nvidia. Just wait. Yeah, I'm sorry, Nvidia. I didn't say, please give me access to GPUs. Please don't take those away from me. Yeah. It's interesting that you look at some of what was being developed for Web3, legitimate. Had use cases, had applications, but there was so much over promise under deliver, just across the board, even where there could have been value.

Eldad: I'm sorry, I apologize.

Vin Vashishta: that technology that had potential was swept aside and all of it's a con. And that's the danger we're in right now.

Eldad: So in 100 years from now, they will look at us and say, oh, it took them so fast to get AI running. Nobody will remember the reports and the dashboards. And everything we've done, it will all be replaced by just a system that works. So that's the next step for everyone. And we're just here to support it, each with its own value in the big data value chain. So thank you for that optimistic outlook for everyone.

Vin Vashishta: We're putting in the pieces. Yeah. We're putting in the pieces and laying the foundation. And if we listen to the right voices, you know, the Joe Reis, if we listen to those people, you're going to have a more successful approach than if we listen to some of the Thread Boys and the people who get way too much attention, media-wise, social media-wise.

It's great to have an announcement. I want to see a product that works. If it, if announcement does not produce product that works, we have to start discounting those people and letting those voices just kind of fade into nowhere. And focus on being the people that build things, being the people that deliver value. There's, there's this tendency of, Oh, we want to make awareness. You know, we've got to show that these people are no, just believe me, they will fade into nowhere.

They just go away. That's the great thing about it is if you don't deliver two or three times, people just ignore you. So all you have to do is focus on delivering and we'll be fine.

Benjamin: Thank you.

Eldad: I'll boom to that. This is like so important. This is so true for everything we do in life and for data sometimes as well.

Benjamin: Awesome. I think that was a great closing statement, Vin. Thank you so much for being on the show today. It was super fun having you. Yeah, and all the best going forward with all of the other companies where you can have impact.

Vin Vashishta: Thank you so much for having me. I appreciate it. And letting me say a few wild things here and there.

Eldad: Thank you, thank you.

Get started with Firebolt

Free credits. Free support. No credit card needed.

Read all the posts

Intrigued? Want to read some more?