The Data Stack Show - 261: Will AI Permanently Disrupt the Bundling and Unbundling Cycle?

Episode Date: September 10, 2025

This week on The Data Stack Show, Eric Dodds and John Wessel explore how AI is reshaping the data industry, focusing on the ongoing cycles of bundling and unbundling within data infrastructure. They discuss the potential for closed ecosystems like Notion to deliver personalized, integrated experiences and examine recent industry moves such as Fivetran’s acquisitions. The conversation also highlights the challenges faced by both startups and incumbents, the influence of enterprise customers on product development, and the enduring importance of trade-offs when choosing between bundled and unbundled solutions. Key takeaways include the complexity of implementing AI across platforms, the likelihood that market cycles will persist despite technological advances, and the need for organizations to carefully weigh integration, flexibility, and long-term risk when adopting new data tools.Highlights from this week’s conversation include:AI’s Value and Early Ecosystem Integration (1:11)Closed Ecosystems and AI Opportunities (3:21)Personalized Software and the Blank Page Problem (6:17)Transition to Data Industry: Bundling Trends (9:56)Market Cycles and AI’s Role in Bundling (12:56)Incumbents, Innovation, and AI Layering (15:53Longevity of Legacy Systems and Ecosystem Risks (17:56)Switching Costs and Incumbent Advantages (20:33)People Dynamics and the Startup-to-Incumbent Arc (22:50)Enterprise Data Infrastructure: Engineering Challenges (26:33)Fragmentation, Bundling Value, and AI’s Insulation Effect (29:54)Too Many Tools: The Real Meaning Behind Bundling Demand (31:36)Trade-offs in Bundling, Unbundling, and AI (33:40)Final Thoughts and Takeaways (34:34)The Data Stack Show is a weekly podcast powered by RudderStack, customer data infrastructure that enables you to deliver real-time customer event data everywhere it’s needed to power smarter decisions and better customer experiences. Each week, we’ll talk to data engineers, analysts, and data scientists about their experience around building and maintaining data infrastructure, delivering data and data products, and driving better outcomes across their businesses with data.RudderStack helps businesses make the most out of their customer data while ensuring data privacy and security. To learn more about RudderStack visit rudderstack.com.

Hi, I'm Eric Dots. And I'm John Wessel. Welcome to The Datastack Show. The Datastack Show is a podcast where we talk about the technical, business, and human challenges involved in data work. Join our casual conversations with innovators and data professionals to learn about new data technologies and how data teams are run at top companies. Before we dig into today's data. episode, we want to give a huge thanks to our presenting sponsor, RudderSack. They give us the equipment and time to do this show week in, week out, and provide you the valuable

content. RudderSack provides customer data infrastructure and is used by the world's most innovative companies to collect, transform, and deliver their event data wherever it's needed all in real time. You can learn more at RudderSack.com. Welcome back to the Datasack show. This is Eric Dodds and John Wessel. I'm excited to be back from Greece and back on the show consistently. So John, thanks again for having me as a special guest guest here. Welcome back, special guest host. Okay, we made an interesting call. We asked Brooks if this is okay, but we pushed off a guest by a week because there's a topic that I've been chewing on so much when it comes to the way that AI is impacting the data space and things in general.

And so I, of course, called Brooks and said, hey, I just want to pick John's brain about something. Can we record it? Maybe this is just an excuse for us to have someone who's not smarter than us in the room so that I'm less self-conscious about, you know, my ideas. But, okay, so this is a special edition of the DataSack show. And we're going to talk about bundling and unbundling. in the data space in the age of AI.

So are you ready, John? Yeah, okay. Let's talk about bundling and unbundling. So I'm going to present some theories, and so this is going to be, I'm just going to present a lot of stuff, and you're going to tell me if it's dumb or not, okay? This is our new segment, dumb or not.

Dumb or dumber? Okay, I've been thinking a lot about a number of things. So AI is very expensive, still. and AI has been, it's shown itself to be very valuable in really specific areas, disciplines, tasks, right? So, you know, IDE's is an obvious, you know, is an obvious area.

You know, I think about companies like Hex, you know, who are leveraging AI to do stuff, you know, in analytics for people working with data, analytics, data scientists, you know, analytics engineers, et cetera, very effectively. I mean, it's still a long way to go, right? but generally tools where people say, okay, this is a new way of working a new tool in the toolkit.

But broad application of AI across an ecosystem is, we're very early, right? And I was trying to think of an example of this being done really well, and it's pretty hard to think of those. So one of the really obvious ones would be the Google ecosystem. especially as it relates to G-Suite because you have email, calendar, documents, spreadsheets, everything, right? And that still is localized, right? You know, you can sort of generate slides in Google slides now.

You can, you know, have Gemini help you in your Google Doc that you're writing and those sorts of things. But it's still not connected across the entire ecosystem. I actually think Notion is really well poised to make a big difference here. Theirs is still pretty localized just in terms of Notion AI and it sort of pops up in a dock to help you. But they acquired a calendar company, they rolled out an email product, everything under the hood is a database and the core Notion offering. Super interesting. I'm actually, I think we'll see like some incredible strides there. But all of that to say, I think that those closed ecosystems will actually make the biggest advances in terms of broad magical experiences. across an ecosystem as opposed to something really specific

where I'm writing code in an IDE, right? Which I have multiple different applications that are connected in an ecosystem. An AI is sort of taking all of that context and generating something really interesting. So I think those closed ecosystems will probably create the most interesting experiences in the near term, assuming they can figure out

the cost side of it. So, okay, first idea, agree or disagree? on the closed systems yeah just the you know delivering sort of the broad ecosystem based you know right experiences with AI yeah I think company size is a big factor there I do think like the notion example like small and some mid-market companies I can definitely see that being a thing where it's like okay we built a knowledge base in notion we started there and we like it and then we started and then like oh like it has a note taker like I'll try that that out like oh this is pretty good and then like kind of go down that road or the calendar email

etc i think that makes sense i think i mean some of this is a roadmap thing and a style thing interesting thing about notion is like notions roadmap but the interesting thing about notion and tools like it is it's for some businesses even small businesses it's not opinionated enough it's like if you're the type of person that wants to like i just want to do what i like like and i have like a vision of what I want to happen, like the flexibility of notion and like some of the tools like that are really good. But if you're like, I don't really know what to do, like there's some more opinionated software out there that might be better for you. If you're trying to do some CRM type thing or marketing thing or whatever, it's like, I don't like, I don't, like,

I've never even seen a, you know, a database of, you know, CMS database or something. Interesting. And again, what percentage of people are in each bucket? I don't know. I've seen. both, like one of really opinionated people that know exactly what they want and I have a plan, like, Notion, I think is really good for that. But there's a big group of people in the other category, which is like, I want this tool to be like to tell me what to do to like, as far as best practices and what I should be doing. Well, okay, so this is interesting and is it going to lead into my thesis around this for the data world. Yeah. But I think the opportunity for notion as a closed system with your knowledge base,

You know, a lot of, you know, people will keep, like, project, you know, manage projects, you know, keep a customer system, you know, or a list of customers, et cetera. Well, if you have all that connected to your email and calendar, it can actually, like, make, it can guide you on, you don't know what to do.

And, yeah, I agree. Like, it's kind of just a database with a UI layer on top. But what's really interesting, and that's why I think that they have a huge opportunity, is that it's just a database underneath, right? Right. If you connect that with everything, you actually, it's not, there's not a blank page problem because they have enough context to say, okay, here's actually like, hey, I mean, this is

absolutely within the realm of possibility. Like, we built a CRM for you. Right. A personalized CRM. Yeah, a personalized CRM for you. It doesn't have fields and features that you care about. Because it's just the database, right? Yeah.

Right. And then, okay, so we built a CRM for you. and we are going to like when you open your email we're going to display we're going to auto draft all the emails that you need to send to these particular customers or like prospects whatever that is right all those sorts of things i mean really interesting opportunities you know not easy to do yeah i agree there is some blank page problem but i think that's that will be solved that's yeah i think it's i think it's solvable i think it's not i think it's not trivial to solve but I think it's totally agree because the version one is like could you do it now with some

of the AI stuff and just like hey tell me what to do like yeah you could but I think some of this needs to get so simple as like onboarding is like create my personalized CRM and you click a button and it does everything I haven't seen too many experiences like that totally well and there's a little bit of a chicken and egg in that if you already have in order to do that well you already need some existing data, right? Yeah, which is in places like email and cal. Yeah, exactly. And contact. Totally. So if you connect those applications, right? But hold starts pretty hard. Yeah. Yeah, not easy, but I think doable. Well, and it's that theme of like, what are people calling it, like personalized software? Like, there's been a lot in the news recently about that of like all these

companies are going to have this personalized software that like before it's like, well, of course you buy an ERP or you buy a CRM or buy. And now it's like, I don't know. Like, Maybe there's like a platform thing where you kind of build one. Totally. And the benefit of it is not that it's necessarily better or can do more, is that it doesn't have the things you don't need. Like that's like half the benefit. It's like here's a CRM with the 20% of things that I actually care about.

Yep. And then when I onboard people to use it, when I use it every day, it makes me happy because I don't have to trip over the 80% all the time. Yeah. Like there's things like, that makes sense to me. It is going to be fascinating though because I think opinionated software. where it tends to be a better experience. Right.

But a platform like Notion does have the ability to, like, layer in some baseline and then, you know, sort of customize it so you don't get what you don't need. Yeah. Okay. Let's transition into data. So I think, so number one, I think we see bundling, right? Right. I think we're in a bundling stage for sure.

Yes. What are the top examples of bundling that come to mind for you? you recently a lot a lot with five train and snowflake of mine yep because we had all this like we're going to actually five train another example recently right so you had this like we need etl and then you need reverse etl and then you need storage and then you need metrics or modeling like there's just governance is like you know six or eight things and then with five Trin and the obviously has a reverse ETL component now, ETL component through census acquisition for reverse ETL. And then the recent one in the last couple weeks is the SQL mesh.

I can't remember the parent company of that. But Tobias, I think, is the guy who started. Anyways, like the SQL mesh stuff is part of, it's going to be part of the five turn ecosystem. So now they have, oh really? I didn't see that. Yeah. They have extract. Brooks allowed to fact check me on this. But I'm like 98% sure. So they, now we'll have extract, which is what they've always had on extracting and loading as part of their tool. Then they have their own modeling thing, not dependent on DBT. It's like their own thing. I think it still stays open source, but they have their own like thing there. And then they have the reverse ETL part. So they're full sack except for storage. And then like if you told me like next

week that they buy some kind of like storage thing, like I'd believe you. Right. I mean, I think that's like where it's trending. And then Snowflake. is in, I think Databricks to some extent, but for sure, Snowflake is going on the other direction of like, all right, we're going to do, I think it's called OpenFlow. It's now part of the data in and out with OpenFlow. Yep.

They have DBT integrated in preview now where you can like run DBT inside of Snowflake for transformations. And there's probably, I'm sure there's other things I'm forgetting. But like you see that from like two big players like going to where they're essentially going to be both like data stack offerings. Yep. So part of that is, part of that is just a natural cadence of the market where, you know, previously there were

a number of companies that built things that were, let's say, more, you know, this, it was the classic, you know, we're in a VC bubble. These companies are getting funded for things that are just features of larger poppics. Yeah, right. Which, you know, of course, the ground truth is always, you know, more complicated than, more complicated than the narratives that you see in the news. But is proving true, especially when you look at examples like Census being bought by 5Tran, right? Where it's like, okay, this actually does make sense as part of a larger pipeline platform, especially for the enterprise.

There will always be, you know, smaller companies who serve like really specific needs of, you know, of other companies. So part of this, I think, is a natural cadence of bundling in a market cycle with a lot of factors, right? I mean, VC funding is one of them. General economic climate's another one. buying patterns, you know, the whole push around data, you know, especially as it relates to AI. So a number of factors there. But how much of it also do you think is driven specifically

by envisioning a future where you need to bundle in order for AI to work well for certain things? I think that might be the side effect, to be honest. I'm sure, I'm sure for some people that's an act of like, hey, we're doing this for that reason. I think it might just, just be a side effect of like there's just a practical like hey we raise a bunch of money we need to grow like all right we should probably buy somebody because like we're not going to see the growth without that yeah that's just really practical yeah and I think we're going to see that because of like all the majority of the money moving into like AI related companies not that some of these companies you know do have an AI component to it but there's just a lot of money that was in the data

space that you know is now in AI application layer stuff so I think that's just a practical thing, but for the ones that are driven, like, hey, we want this, like, full ecosystem thing, like, how much of that's related to AI? I think the interesting question to me is, like, with this unbundling, bundling cycle that we've seen, like, forever, bundle and you unbundle, you bundle. Like, how does AI impact that? Like, do we, like, it'd be, it's interesting to think, do you, I don't want to say do you ever, but, like, do you ever unbundle again? and if you like if AI wasn't part of the conversation I was like absolutely like these companies get big they get bloated they stop innovating here like somebody's going to come along and disrupt like

they always have and you're going to unbundle again because somebody's going to do a better job of like X Y and Z always happens that way yep I think the interesting question to me it is AI disrupt that cycle and it and I'm really not sure I kind of think probably not because it's a people problem and that you just have like when all the people are together and you have to maintain cohesion with thousands of people on really complex products, which like AI does a lot of cool things, but it doesn't simplify products yet. Like at least by default, like AI is generally like going to come up with a working solution but not necessarily the best or simplest solution. So I think you could potentially end up with like extra, at least for a while extra complexity. Yep. And then you

still have the people problem. Maybe it's less people, right? Because they can work, you know, maybe they can be more efficient with AI tooling. But I still think you have those fundamental problems, which slows innovation and then like makes somebody want to disrupt, you know, one of these like all in one ecosystem. Like, oh, like this is terrible and this is terrible and this is terrible. Ingramatica. Yeah, exactly. Well, okay, let's talk about that. So be, so I want, I love the question of how will AI specifically impact bundling and unbundling, but one really interesting thing to me about informatica, for example, is that the, especially as it relates to data, is that if you think about the challenges

related to those like really large incumbents tend to get unbundled, so Informatica breaks down, you know, 5Trans or whatever, right? So a more model. faster, better user experience, more fully featured, whatever it is, whatever the, you know, whatever the specific characteristics of the unbundling is. But what's interesting with LLMs and with data is that the, and the thing that is a challenge for informatica is that they've been around for a long time. And so retooling and innovating is culturally costly, but literally costly from like an engineering standpoint, right? It's just, it's hard, right? If someone comes in and they use all the latest technology. and so, you know, that's just a difficult thing, generally.

But if you have an ecosystem with all of the data, you can layer on, you know, AI features on top of that, right? It can be done super poorly, you know, which a lot of companies are doing, where it's like, just put AI on this and, you know, can very poorly. But it can be done really well. And what's interesting is that it doesn't require this massive, you know, sort of retooling. or, you know, rewriting from the ground up or whatever, right?

Like, you can actually sort of layer that in. And so part of me wonders if we won't see for companies that can overcome the cultural side of it. If you actually won't see it's like, okay, well, you just have all this data collected in an ecosystem. Well, that's actually a very interesting foundation on which to try some interesting things. Right. Yeah. I think it's funny you bring up Informatica because I was actually just thinking about not them specifically, but the general, what is it called, the Lindy principle, like the principle that states, like, the long, how long something's going to be around is a direct correlation to how long it's been around. So, like, informatics has been around 30 years. Like, it's more likely to be around in 30 years than like some other thing today that may be really hot and great and whatever. Yep. But Informatic was actually more likely to be around. Yep. And it was related to, and it reminds me of bundling and unbundling, because this is a completely different take on that. But really interesting to me is let's talk, like,

in the data space of say you have Python, like smaller company and essentially you have Python scripts that like do most of the work in your company. And they run on a server somewhere. From that principle is like what's the likelihood that there will be Python scripts on a server somewhere in 10 years like running? Not with like all the fancy new like tooling in the sense. So it's like probably fairly high. Not like for one company specifically, but like in general. Yep. Yep. And what I think is really interesting like when considering like future when considering risks, is like that principle of like okay like oh we got to like move to this latest like platform we got to use this we got like put all this stuff in this new thing and then it's like oh they got acquired and they're shutting the product like you know just like things that happen in our space just so interesting to me with the bundling and unbundling thing of like there is still going to be this class of things well well I think we'll probably still build her own Python scripts on servers in 10 years but like any of this other stuff like I don't know sure so relaying that back to like bundling, unbundling thing,

in the closed ecosystem thing, like, I think there's just, I just think there's going to be, it's going to be cool, but I think there's going to be some risk there where you're like, all right, like, I'm all in on whatever the like bundled thing is. And certainly like at some size,

like those things aren't just going to like go away overnight, like they're publicly traded or they're whatever. But you do end up like getting tied into an ecosystem. system. Sure. And, like, you have to be able to find skill sets that, like, know that ecosystem. And if it's, you know, let's think about, like, I don't know, let's talk like Cisco or something. Like, people that, like, are Cisco certified are, like, pretty expensive generally. Yeah. So you just have this, like, interesting, like, yeah, world where, like, I don't think people fully consider, like, the, all the implications of, like, kind of going all in and, you know, in a tooling or ecosystem like that. I totally agree, which is a much more articulate way of making the point that I was trying to make, which, or that I was making my way towards fumbling my way. But if you think about informatic or any other large sort of bundled system, right? I agree.

You've achieved immense distribution. The switching costs are phenomenally high, right? generally the systems are driving core business logic generally they are like you know i say tightly managed not that you know most company's data is really messy right right right but tightly managed in that they drive core business processes they have like clear ownership they have you know that's sort of baked into how the company uses the system right right and so even if there's sort of this fundamental disruption happening, there's still a lot of time there because of those dynamics, which I think creates this interesting world in which those companies have a lot more,

I mean, this is ironic to say in the world of AI, but they have a lot more time. Yeah, right? Because the critical nature of the data and, you know, sort of driving, you know, core business logic, infrastructure, KPI, whatever those things are. And so it's like, okay, well, if they can actually figure out how to integrate AI, they can be in a place where a customer is going to say, well, great, like you have these AI features that we saw in this competitor, and so we're not going to switch, right? Or you're building it, you know, and it's good. You know, like, that's a huge opportunity for the incumbents to sort of ward off disruption, I think, especially in the data space, as compared with something like, I mean, HR software, where like an LLM is

automatically reviewing resumes or applications and like summarizing it, right? It's like, okay, well, disruption is going to happen so quickly there because you can just, those are essentially, you can easily switch that out, right? As long as you can get your candid data out. Right. So that is, I don't know, that's going to be interesting to see, right? Because usually you see the incumbents is sort of like, okay, they have distribution and they have a moat with those things and switching costs are high, but they sort of like slowly, you know, sort of slowly fade into the background. But that may not be the case. Well, here's the other interesting thing about this conversation,

which is also a people thing, is if you've got this, traditionally, you have this arc of successful software or data startup. Like, you know, have the idea, start in some kind of niche, raise funding, do the whole thing, expand from that niche, grow, like, then IPO, like, you know, some kind of, let's just think of like an ultimate success story

of what people would ultimately want here. And then you've got this arc. And the interesting part of the arc to me, and I wonder how AI impacts this, is at some point, the people that are the startup people that help start that company are no longer part of the company. Every time, like almost never does any of those people stay until like IPO. Maybe the founder stays until IPO. Rare. But like after IPO, like they're gone. They're not staying. So then essentially you have a whole different group of people that have no idea like why it was started. They don't understand any of the struggles of starting a company. They don't understand the core problems sometimes. anymore. So that's typical what happens. And then you have, so then you're at that point and then say you're consolidated. You're this nice consolidated ecosystem at that point. And then like if you look at the company profile, there's just a different person that wants to work at a 5,000 person publicly traded company and data than wants to work at a startup and data. I don't think that's AI doesn't change that. So I think the big question then is like back to the bundling and unbundling and

disruption is like because I don't see any change in that pattern yet like why would unbundling and bundling not continue because like if it's going to follow that pattern with the people like the software is ultimately a reflection of the people building it to some extent do you end up with another unbundling in five years because it's like oh like we've got all these ecosystems and this one's efficient here into fishing or maybe AI is so good where the you know the tech that can get cleaned up fast because like AI can do it or the you know the innovation is more done by the AI than the people that would change the equation yep but other than those types of things like I don't know if that equation changes yeah it's man what an interesting thought we're going to take

a quick break from the episode to talk about our sponsor rudder stack now I could say a bunch of nice things as if I found a fancy new tool but john has been implementing rudder stack for over half a decade. John, you work with customer event data every day and you know how hard it can be to make sure that data is clean and then to stream it everywhere it needs to go. Yeah, Eric, as you know, customer data can get messy. And if you've ever seen a tag manager, you know how messy it can get. So Rutterstack has really been one of my team's secret weapons. We can collect and standardize data from anywhere, web, mobile, even server side, and then send it to our downstream tools. Now, rumor has it that you have implemented the longest running production instance of Rutterstack at six years in going.

Yes, I can confirm that. And one of the reasons we picked Rutterstack was that it does not store the data and we can live stream data to our downstream tools. One of the things about the implementation that has been so common over all the years and with so many Rutterstack customers is that it wasn't a wholesale replacement of your stack. it fit right into your existing tool set. Yeah, and even with technical tools, Eric, things like Kafka or PubSub, but you don't have to have all that complicated customer data infrastructure.

Well, if you need to stream clean customer data to your entire stack, including your data infrastructure tools, head over to rudderstack.com to learn more. Okay, so circling back to does AI change that equation, which you brought up earlier. It's a fascinating question. And I think the data

space and let's use the term enterprise data infrastructure, you know, quote, but it's really hard. It's really hard, right? I mean, yes, there are a lot of like ways that you can stream event data. Yes, there are a lot of ways you can like ETL data, you know, but if you look at the broader landscape, like the companies that, of all the companies that were, you know, started in the last, let's say, 10 years in that space, like the ones that became very large, it's a small number, even though it's a huge market. And part of that is because it's really hard. Like, it's actually like the engineering challenges are like particularly difficult, right? Yeah. As opposed to say, and I'm not diminishing the software engineering effort, let's go back

to the HR system example, right? Yeah, I mean, they're like difficult things about that. But they're just some things specific to the nature of like moving enterprise data at scale that are like particularly And here's one of the thing that I failed to mention on this arc that I think is really important, because I'm just talking about internal. The other thing that, like, is obvious, probably, but, like, is worth stating is if you're going to, like, quote, make it, like, go all the way through the cycle, IPO and stuff, guess who your customers are? You have to have enterprise customers to do that. Like, in this space, like, almost 100% of the time. Like, you have, and guess what? Enterprise customers are going to impact your roadmap and, like, what you work on, for sure.

and then like that also impacts the people that want to work on enterprise things so there's this and it's necessary and important but there's a slowing down if you're going to work with like fortune 100s you cannot go at the same speed and i don't think AI is like quote fix changed this yet yep like there's a slowing down of like hey we care deeply about security we care deeply about data governance yep so there's a slowing down motion you're going to see in any of these companies when you get up to that like altitude or speed or velocity whatever you want to call it

and it's just going to slow down by the nature not just to the people internal but the customers which totally that's the bigger that's probably even bigger influence than the people yeah I totally agree with that because I think you know with as as LLMs which this is probably a separate episode

unto itself maybe we'll do one of these where I just hijack the episode with that you know because I feel I'm excited to be back, you know. But I think in a lot of areas where the, like, underlying software, I mean, let's just talk about, you know, processing billions of data points in a short amount of time, unbelievably difficult engineering problem, right? Yeah. I could vibe code an HR application, you know, that's backed by Superbase, that, you know,

all of those things. Like, you could do that in a weekend and, like, some people would use it and theoretically pay for it if it's like, you know, solve some sort of problem, right? Right. Yeah, but even in that scenario, like if you want to sell it to the enterprise, you need governance and security interests down together.

Like you couldn't, but to your point, like, if the, like, the, you know, processing billions of records, like, cool. Like, you know. Totally. Like, if you spend enough money,

like, you could probably figure out a way to do it. But you can't do it efficient. Right. Yeah, exactly. You know, yeah. And so I do think what's interesting about the data space is that, you know, there,

is fragmentation happening with people, you know, just building data tooling at a much lower cost than they were able to do before, you know, because of the way the ELMs have impacted engineering and all of the infrastructure that's being built around that. However, my hypothesis is that the disruption that comes from fragmentation will actually impact the data space less quickly than it will other spaces because of the nature of the difficult problems. because bundling, because there's so much more value in bundling, right? Right. I mean, the value from bundling in data, so you brought up 5Tran, right? So you're pulling the data, you now have a modeling layer, and then now you can get that data back out. What's possible with AI there is largely possible because of the context that comes from bundling, right? And so the, it's, the fragmentation is going to happen, I think, in very niche spaces, which is fine. And it's like, I think there's going to be a whole. industry around those very specific things, right? But what do you think? I think that relative to

some other industries or spaces, like I think the data space, especially in, you know, let's say like upper mid-market enterprise, will probably be slightly insulated, slightly more insulated from fragmentation due to AI. Yeah. Especially if you're bundling. Yeah. I think that's true. I think there's another like component here too with the bundling on bundling. of I hear from a lot of people that this is what they say and I'll tell you what I think they mean they'll say like oh like we have too many tools great lead in by the way yeah this is what they say I'll tell you what I think we have too many tools like I just want one thing to look at I just want one thing we have too many tools yeah and then the and then the sales pitch that I was like oh great

look at this thing like we now can do all these things like you just have to buy us and like it's taking care you don't have to have multiple invoices or learn multiple tools like that that's a big sales pitch especially right now I mean that's a huge driver behind the bundling cycle generally. Yeah, right, right. And I get that. But, like, I think people do mean that. They really do just want, like, one bill and one thing to me, whatever. But I think what they actually mean is, like, hey, I want all this to work together. I want you to be responsible. And which is fine. Yeah. And I think, I mean, I've been a part of teams where we've used, like, a pretty good number of tools that all worked really well together. It was a great experience and

awesome. I've also been a part of teams where, you know, the opposite of that is true. And I've been a part of teams where you're used unified data stacks. We're all in one vendor all supposed to work together because the vendor's responsible and they don't work together and it's awful. Yep. That's the thing that people miss, especially at like an executive level of like, oh yeah, well, we just went all in with like these guys and like we bought their thing. And like it'll work together. I mean, it's all the same brand. It's like the brands match. It's like those are real people on real separate teams and they have to create interfaces between the things and that is like not necessarily done well. So they might as well be different vendors

because it's such a big company and the products are so separate that like sure they're supposed to work together but they don't. Yeah. Or they don't work well. Yeah. Yeah. So I think that's just like a misnomer. Yeah. People think that by like delegating that like responsibility to the vendor of like oh yeah, let's all your stuff. You'll like it's like no, that doesn't necessarily solve any problem. And AI is not a solve for that. Not yet. Not yet. Yeah. Well, I mean, well, maybe that's... Well, it's not, actually, because it's not in your control. Exactly.

You have to depend on the vendor to like to do it. Right, right, totally. The vendor, yeah, and then, you know, there's all sorts of custom business logic. But, yeah, I think that's a really interesting... I guess we got to lay in the plane here because this is a... We started recording late, but my fault, because I hijacked. But I think that's an interesting thing is when we think about bundling and unbundling, even when you consider AI as part of that, ultimately it comes down to tradeoffs.

Yeah. Right? And I think that's actually one of the things, and maybe like a huge misnomer of like AI being a promise of you can sort of do whatever you want, a huge misnomer, right? Where it's like, okay, well, if you use an all in one system, there are tradeoffs, right? If you stitch something together, there are tradeoffs. If you build it yourself, like, there are tradeoffs. And we're definitely not at a point where AI, like, dramatically minimizes those to the point where, you know, the lines between those are getting blurry, which is super. interesting. All right. Well, good one. Yes. Bundling and unbundling. Man, we could keep going.

But all right, we'll call it because we're at the buzzer. We got some great guests coming up. So subscribe. We'll notify you on your podcast app of choice. And we'll catch you on the flip side. See ya. The Datastack show is brought to you by Rudderstack. Learn more at rudderstack.com. Thank you.