Latest / ThinkData Podcast / S4 | E13 | Why Data Access Will Decide the Future of AI with Anjan - CPO @ Fivetran

Transcript

S4 | E13 | Why Data Access Will Decide the Future of AI with Anjan - CPO @ Fivetran · ThinkData Podcast

• 0:03
  Welcome to the Think Data podcast brought to you in partnership with Mydataworks. If you want to stay up to date with the latest breakthroughs and trends in the world of data and artificial intelligence, and if you're curious about some of the strategies that companies and founders use to launch data and AI products, then you're in the right place. Our aim is to bring together a diverse lineup of fantastic guests from the founders, through to accomplished leaders and product owners at some of the most fascinating data and AI companies worldwide. They will each offer you their own unique insight into what it takes to launch and scale a great data business. Thanks for tuning in and I hope you enjoy the episode. Welcome to the Think Data podcast and today's guest is Anjan. He is the chief product officer at Fivetran. Everyone's heard of Fivetran, but simply put, they are the company that is at the heart of every modern data stack. In this episode, we're going to dig into who really controls customer data, what data ownership kind of means in this new age of AI, and how things like API restrictions are starting to shape what companies can and can't build. We're also going to explore what happens to innovation, which is a really big point here, if data becomes harder to access and how Fivetran are leading the way in thinking about data freedom in this new AI world. Yeah, it's really good to have you on, Andrew. And I know prior to this, you know, you spent, you've been at Fivetran a couple of years now, before that you were leading product at Precisely. But, you know, Fivetran, obviously one of the leading lights in this kind of modern data space, but what drew you to the organization first and foremost? And secondary, what is your role there? And ultimately, what are you leading?
• 2:00
  Thank you, Alex, for having me on the podcast.
• 2:03
  Pleasure.
• 2:05
  FIFTRAN, as you kind of stated, is like a very important part of the modern data stack, is the leading product that disrupted how to move data. And I've been in the data space for a long time and I've come across FIFTRAN. I was really, really impressed with the product. the vision, the first principles, thinking how it works and in the end, like delighting customers. So when George and the team here pinged me and I was like, yes, I know the product and we kind of gelled the culture at Fivetran is really amazing with the customer first and sort of always thinking about how to innovate to provide more value. So was a big fan of the product, continue to be a big fan of the product. So product is as a product person, when you're delighted to, you're always delighted to work on something. that you love. So this is like one of those opportunities that I could pass on. Yeah. So that's kind of the reason I joined Fivetran, Alex.
• 2:52
  Interesting. And from a product standpoint, obviously when you get to a size and kind of position at Fivetran, find themselves in obviously trying to stay ahead of the competition and keep moving forward, ultimately falls on the kind of shoulders of the product function. So what is your day to day or week to week look like typically?
• 3:14
  Yeah, it's a good question. I think we, as a product org, kind of lead the product org here as a CPO, we want to listen to our customers first and foremost. We kind of want to make sure that we are pushing the envelope and we are thinking very deeply about the problems that they solve. And then you kind of translate that into what we need to go do. And so one, I always kind of tell like as a product manager, you want to deeply understand the problem and then inspire the team here to go drive it. Thankfully, like we've got a set of customers and we count like OpenAI as one of our customers. We get to see very, very interesting use cases. We get to see when they are pushing the envelope. And in some companies, they kind of look at that as like, oh, that's like an edge case. We're not going to do that. At Fivetran, we kind of embrace that complexity. We embrace like it doesn't matter. Like every connector needs to work the same. It doesn't even if like very small number of customers are using it. Or customers kind of really blowing past kind of what it means from throughput or latency point of view. Those are the same. And that's kind of ethos of Fivetran. And like as the product team, it's kind of built around that. It's like nothing's kind of like if something doesn't work to standard, it's a bug. And it doesn't matter like where it is. And we have like an interesting problem where we support 700 connectors. And that number is kind of growing very rapidly with our AI-generated connector. and it doesn't matter there's a lot of incidental complexity that is in those SaaS applications databases and so on and we in other companies they're like yeah okay well there's only so many you can do we embrace the long tail and and I think the problem is like inherently kind of driving that innovation and mode so that's kind of on the core side and then we'll talk a little bit about ODI and we're pushing the frontier in terms of data lakes in terms of like building connectors The upcoming merger with DBT, they've sort of done this sort of similar narrative from the transformation side. So we're very excited that us coming together, we provide like a data foundation for AI going forward.
• 5:13
  Yeah, it's really interesting. As a recruiter in data for about 10, 15 years now, I've certainly seen over the last, and obviously it does coincide, obviously, the hype around AI specifically. people. I have seen a fundamental change in the way that modern data teams are built, staffed, that we're almost seeing this kind of wholesale market shift of we're almost a victim of our own innovation and efficiencies here. And obviously, Fivetran being so central to that modern day analytics engineering, you know, kind of the ETL piece, everything around data is changing. But what's your kind of view on that? because there's a lot of people who listen to this who let's be honest the market within data analytics is pretty tricky for people right now and everyone's blaming ai and saying it's just ai but actually you know i'm assuming five should have a different perspective on that and we also want to drive this change for the community uh yeah see where this goes no i think it's it's interesting because obviously on one hand you've got ai generating so much data And yet AI is also relying on these kind of models, sort of on these data environments to kind of power its own model. So I'm seeing, you know, are you beginning to see this kind of wholesale shift in terms of how companies are accessing and moving their data and kind of what impacts that now having down further along?
• 6:37
  Yeah, we are seeing that. We are seeing that. I mean, in some interesting way, it's back to the basics. Like, hey, make sure you have your data. foundation right, make sure you're moving all your data in one place, make sure you kind of have a storage layer. So you are seeing companies realize now like, okay, like in the last three months, this became way more important than even last year. So doing the basics right is going to hurt you because if you don't do that, then you're going to spend a lot more money on token costs. You're not going to get access to some of this information. You kind of need to figure out how you create this architecture where employees and agents can thrive. So we are seeing then all of a sudden that customers are like, oh, I didn't need to connect my JIRA before. Well, now I need to connect, or I didn't need to connect the system that's proprietary that had some information. But guess what? That had some really important context that your agent can now process. We are seeing more unstructured information get processed through. And there's a lot of unstructured data, even in sources like in Salesforce or Workday or whatnot. So we're seeing customers pull more data out. the sources that they already do. We're seeing customers move new data that they haven't done. And the third sort of category is, this is very early. I suspect it'll grow. You're seeing customers create sort of these ad hoc applications that they haven't before, right? Fancy vibe coding apps. And that's a new trend. I suspect you're going to see more of that. So that's going to now create data. Well, you can't have like hundreds of these apps not connected, then you're going to lose the value. You still want to talk to these transactional applications that you create. And all of them have very important context. So it's like, I think the value of data and data foundations just exponentially increased in the last few months because now you could do a lot more. The things that you couldn't do before, you can now. And you're going to have to work overnight and on your direction and do really, really incredible things.
• 8:32
  Yeah. And I also think as well, from the conversations we've had, a lot of organizations, let's be honest, everyone's trying to move towards more agentic AI and people are trying to. harness whether it's open source or their own proprietary models but they've also realized if they've not got their data environment set up correctly then they're going to invest and waste a huge amount of time and energy into building something that ultimately isn't fit for purpose but it does beg the question though who owns the data you know all this data has been kicked out by chat gpt claw everything that's being generated what's your honest view on kind of who owns that data and in the world of AI How are we going to put some guardrails in place to ensure that people's data is ultimately theirs?
• 9:17
  This is a very important topic, Alex. I think customers principally, first and foremost, own their data. A lot of us vendors, whether you're on the source side, destination side, or data infrastructure companies, we are custodians of customers' data. We move it or we create it. It is silly to say that, hey, I now have been using you for... Five years to put my sales data and now I have to pay you because I've kind of used your system of record or financial data or whatnot. I think that's a terrible, terrible thing. Now, the patterns at which you use your data, like a human would go in and enter sales information or you've had APIs. But now you're going to go create. a dashboard for every human. Like I might go on demand, go create my own sales dashboard or my agent might. And the interaction pattern has changed. All of a sudden, just because we've used you for five years, you can't say now you're going to charge 10 times more. That is crazy. And I think that would be, companies that do that are going to be under threat medium to long term because those are the ones that I've heard CDOs now say, look, I own my data. My data is way more important. I need to innovate with it. And it's critical. that the entire ecosystem, all the way from system of records to us, enable that and say, hey, we're going to do that. So we're going to take a very strong position on this. We are going to be on the side of openness. We've published a benchmark called Open Data Infrastructure. You can go to opendateinfrastructure.com. And we're going to be like very simple, very simple algorithm on, if you're a system of record, how open are you? Do you provide an API? Do you create, sometimes you can provide an API, but create all kinds of tricks. Say, oh, you can only take this much data out or you can rate limit or you don't provide a change record when data changes or you don't provide when deletes happen. Right. Those are all tricks. They kind of like on marketing purposes works. And there's some vendors who do a good job and kind of go to that website and see. But we think that certainly the value of data has increased, partly because customers can go do more with it and you can't lock that down. And I think companies that embrace that are going to win in the long term. You might create some short-term... sort of gating. And then like that is only going to hurt. We could do the same thing. We could say, oh, now we're going to create sort of mechanisms to kind of prevent that. And no, we're going to embrace it. We're going to create more open capabilities in this open data infrastructure. It's us and our competitors even, like if our competitors want to do things with this open data infrastructure, we would invite them. We're going to embrace that. We think an ecosystem now needs to be more open because we're going to see an explosion of usage and we're going to push very hard. And we're seeing customers complain about this, and we're going to champion the openness with our customers on this.
• 12:01
  That's huge. You know, I think companies like Fivetran lead in the way. You know, such a big, like you talked about at the time with DBT. I think these organizations are well-respected, well-regarded. And if you're leading the way, then ultimately people will hopefully follow suit. Because in your opinion then, if API becomes more restricted and we're talking about companies that are trying to build AI products, what ultimately would happen? Are we going to see companies kind of, again, not being as open or honest with what they're doing with their data? Or is it just going to affect, is it going to slow down innovation ultimately?
• 12:34
  My hope is companies that are open will thrive, whether that's a system of record or someone that's providing analytics or whatever you want. And companies that don't eventually will get disrupted because someone's going to provide like an absolute agent first, agent friendly architecture and agent friendly. data access policies. And we think that's the world we're in because the return on investment is too high and the value is too high. Where in a few months, you might see the capabilities of models get even better. So I don't think you want to prevent that innovation from taking place. And you can't. And the pace at which AI is going, we think openness is going to win out. So I don't think closed architectures are going, I think closed architectures had a point in time where they were not successful. Now, I think it's open architectures that are going to win for sure.
• 13:26
  Interesting. So when you're, whether it's on a direct business relationship or just through the network that you're operating, if you're talking to, you know, those kind of CDOs or other organizations around data freedom and obviously this open API and open data environment, what's kind of your number one kind of bit of advice to them or suggestion to put them in the right kind of mind frame?
• 13:50
  Yeah, it's sort of been two kinds of things. There are. We are seeing more awareness on this. So they're like, I think even in the last few months, this issue has kind of gained prominence. And that's glad that more folks are talking about. We're talking about it like others are talking about it. So I think a few months ago, this was like less well known, but now they know. I think we are encouraging data leaders to look at their contracts when things come for renewal, push for that access. Then like also take first principles thinking on why. So when you kind of go explain the details and the architecture, like it is, they're embracing this openness and they kind of understand that like, hey, it's their data and they kind of need to own it. And at that level, it's like super high alignment where we are trying to help customers is like, OK, well, here's all the ways where you can kind of get tripped up and we're going to publicly evangelize those sort of bottlenecks and how they can kind of navigate.
• 14:47
  I want to answer this question broadly speaking about kind of AI and where we are on this curve, because obviously you're talking about, you know, you can be spot on in terms of how powerful these models are becoming and obviously different reiterations and versions that are coming out. imminently what impact is that going to have on data teams and broader companies because obviously if we if we're going all in on a genetic ai and we're having all these autonomous agents running what do companies need to think about when it comes to their data environment and their data structure whether it's a team or just internally what's their priority here it's a good question i'll kind of frame some of our own experience here first i i think what this
• 15:30
  unlocks is the bar for a BI dashboard was quite high, like a few months ago. We kind of had to like have a lot of people look at it, like our head of analytics would say, like, okay, what's the purpose of the dashboard? You got to maintain it, curate it. And the bar, the marginal cost of that has plummeted, right? Because now I can go and ask also kinds of interesting questions. So the cost of a question has come down, the cost of query or sort of access to query has come down, not necessarily cost. access to queries come down. And we've seen massive explosion of folks running AI generated queries in our company. Predict that like, okay, well, is that going to be 5X or 10X? And I think these questions are valid. Like customers, like our internal analysts or internal business folks or salespeople or product managers, they want to know these questions to serve a customer or solve a business problem. To give an example, like a product manager to build a feature, he went through all gong calls and like... got all the data out and figured out where the churn risk was in like a couple of days that would have taken using agents. And that would have taken him multiple months. And he wouldn't have even gone and done this. He would have said, this is too hard. I'm going to like do some keyword search. I'm going to do it. But now a new class of questions are now possible. So I think this is great because he's now way more equipped to kind of deal with that problem. Now, this is going to see an explosion of compute. It's going to see an explosion of complexity in terms of architecture. So if you have a very good data foundation and that is kind of optimized for cost, you're going to thrive in this environment. So you can go and tell your customers, which are like if you're a data leader, your customers are internal, you could say, hey, I'm now giving you a lot more value. And I think this kind of helps that earlier question you asked about sort of like jobs. So the internal data team can go, hey, to the CEO or whoever and say, look, our value has gone up because we're now giving you so much more. We can give you. We can give you insight. We can give you dashboards. We can give you applications. We can kind of provide all of this before and look at our return on investment. Nothing's changed about our business overnight. We just can give you a lot more. I don't think, I think companies, so some companies will look at this from a cost point of view. I think the smart ones are going to look at this from a return point of view. It's like, hey, do I want all my salespeople with all of this insight that they didn't have before? Do I want my customer service? folks with this insight, do I, am I going to go after these long tail of leads that I've never could go before? Now I could go and mine all of these leads and go get those customers. So I think it's going to change the business equation. And once customers are realizing that, they'll realize that I think the analytics function, there's certainly like, yeah, I can do a lot of this, but the companies that are thinking about this the right way is like a return on relation. Like now, if there's an analytics team that's saying, look, I'm going to do exactly the same thing I did like a year ago. Well. that is not a good statement. Yes. Then do it for less. But if there's an analytics team that's saying, I'm going to do three times more or five times more. I think there's a lot more, a lot of CEOs out there are going to say, that's great. I'm going to take that because you're going to give me a lot of downstream value on my business objectives.
• 18:39
  Yeah. Yeah. And I, I couldn't agree more. It's such a good, good point you're raising, you know, whilst people always want to hear. teams are hiring are growing there's more opportunity actually it's about what we're seeing now as a wholesale shift there's more companies hiring these folk and actually you're just becoming more specialists in driving more value and being more efficient because of the support you're getting from ai it's not that there are sectors that are going to be disrupted with ai but i genuinely stand by this belief uh if you're if you're a good engineer analytics data behind whatever that is, and you're harnessing and using. some form of AI, your efficiencies and values are going to rock it because you're ultimately future-proofing yourself.
• 19:21
  Absolutely. And that's what you're saying is absolutely true for at a personal level. Our teams, if you're a data leader and you want to transform your team, that's absolutely true because in your next board meeting, you're going to look very good because you've returned so much to your company and your shareholders.
• 19:36
  Yeah, exactly. You're not just going to hire 20 data engineers. Instead, you invested in a product and a solution that's you that's getting more out of your currency. And Arjun, look, it's been really interesting. I feel we could talk forever about this. I think what's really interesting, Fivetran, you know, people will see, unless you've been living on a rock, you know who you are. Check out Fivetran, you know, this data freedom piece, the open APIs, this kind of making data accessible to everyone is just is the way the market should be going. And I hope everyone follows suit. And I thank you so much for coming on. It's been really interesting this morning.
• 20:09
  Thank you, Alex. I enjoyed it.
• 20:11
  Thanks very much.