Did Klarna really delete 1200 SaaS apps?

Over the last year or so, Klarna has been on a rampage to automate away. They've sliced tools, costs and staff. Begging the question...

Is this just a VC narrative or is there more than meets the eye?

• (00:00) - Introduction
• (02:44) - Klarna's Journey and Financials
• (05:49) - AI's Role in Cost Reduction
• (07:51) - Customer Service Innovations
• (10:28) - Internal Knowledge Management
• (13:47) - Introduction to Data Consolidation
• (14:29) - Enterprise Search and Internal Tools
• (15:20) - Klarna's Knowledge Graph and AI Integration
• (16:38) - Deprecating Salesforce and Workflow Changes
• (20:05) - HR and Employee Feedback Innovations
• (25:51) - Build vs. Buy Debate

Never miss a new episode, join our newsletter on revenueformula.substack.com

Creators and Guests

Host

Mikkel Plaehn

Marketing leader & b2b saas nerd

Host

Toni Hohlbein

2x exited CRO | 1x Founder | Podcast Host

What is The Revenue Formula?

This podcast is about scaling tech startups.

Hosted by Toni Hohlbein & Raul Porojan, together they look at the full funnel.

With a combined 20 years of experience in B2B SaaS and 3 exits, they discuss growing pains, challenges and opportunities they’ve faced. Whether you're working in RevOps, sales, operations, finance or marketing - if you care about revenue, you'll care about this podcast.

If there’s one thing they hate, it’s talk. We know, it’s a bit of an oxymoron. But execution and focus is the key - that’s why each episode is designed to give 1-2 very concrete takeaways.

[00:00:00] Toni: I wonder is a tool on top to solve this, just a lovable bolt.
[00:00:07] Query prompt away is it gonna be so, so super fucking difficult to prompt your way to an interface that a human then can interact with?
[00:00:15] Right. But. Until we have solved that, I think it's, it's a pretty fast stretch to kind of replace all of those SaaS tools.
[00:00:22] Mikkel: I think the jury is still out there. Also because it is early days, like hearing story about some of these solutions exposing API keys, which is not a great thing. We're hearing stories of folks who build their own little SaaS company with zero coding experience and now can't scale the foundation. It breaks bugs everywhere because they don't know infrastructure building AI software is like 20% AI and 80% like engineering. You still need to
[00:00:47] Toni: No, no, no, no. It's, it's the quote is much more funny than that. It's it's 5% AI and a hundred percent
[00:00:54] Mikkel: Oh, yeah, yeah. Okay. Yeah.
[00:00:55] if you wanna actually work with it, uh, go try Midjourney.
[00:00:58] It's, it's a whole lot better.
[00:00:59] Anyway, what I wanted to do is just, I took your profile picture and I just gave it some very factual correct information about who you are. And kind of, you know, your experience, where you're from, that kind of stuff. And basically what it told me. Was to go F myself. That's what it told me.
[00:01:24] It's like, ah, I can't create that kind of image where we kind of glorify controversial historical figures, symbols, or regimes, including those started with German Army leadership during World War ii. I didn't say that. I didn't tell it to do anything on World War ii. It just read between the lines, which is
[00:01:44] just effing amazing
[00:01:46] effing.
[00:01:47] Toni: You should do it. You should do it on grok and, you
[00:01:49] Mikkel: They won't care.
[00:01:50] Toni: sure it's gonna work out
[00:01:51] What, what I saw was like you know, which media outlet should you trust more the New York Times or your X Feed, and then. And that was basically like the New York Times is shit for those reasons and you can't trust it, blah, blah. The opinion of the people that's on acts, you only get it on
[00:02:09] Mikkel: No, it's like firsthand factual information you're
[00:02:12] getting. What could possibly go wrong, like I'm just saying, finally we can figure out what happens at area, what is it, area nine 16, whatever area 51. It's like I need to go on X and fact check myself.
[00:02:25] But yeah, you're right. I think you're right. Anyway, we're gonna talk a little bit about AI today because there's a company that's basically just slashing everything and just, you know, blo clubbing in ai and they're. Doing amazing. I dunno if they're doing amazing, but you know, we're gonna talk
[00:02:39] about
[00:02:39] Toni: So I think you guys have probably heard about this one before.
[00:02:44] Um, we're talking about Klarna today. Klarna iss a payment provider and I think kind of a buy now pay later kind of guys, right? I think at the height of. Their storied, they were valued at $45 billion, privately held $45 billion. And now, I think, I wanna say four or five years later after that valuation, they're now aiming to go public between 15 and 20 billion, which is, you know, still insanely fantastic, but it just feels.
[00:03:13] You only got half the way to 45 billion. Right? It's, it's pretty, it's pretty nuts actually. It's like how, how to poop on a 50 billion, a 15 billion evaluation. It's like, say that they were recently, you know, before that varied higher anyway.
[00:03:25] Mikkel: just feel like, honestly, this is a point where we can finally blame finance. They could have moved a lot faster with this IPO stuff, so they could have materialized that 45 billion. Just saying, I think this is probably finance's
[00:03:36] fault.
[00:03:36] Toni: yeah, time was the problem for the 45 billion onto materialize. And then, you know, I think one. And, you know, this is obviously, and you know, we will continue this story throughout the, the conversation. You'll probably kind of get to it in the end if, if we are disciplined enough. Not sure about that Michel, but the you know, they wanted to go public now for a while.
[00:03:54] Mikkel: for the last two, three years, I think at least.
[00:03:56] Toni: yeah, and it's, it's not like, so I think now it's actually, it's gonna happen in 25. They're kind of doing all of that stuff is kind of gearing up. But. Whenever you want to go public like you and I, we know
[00:04:08] Mikkel: Yeah. We've done it so many
[00:04:09] Toni: It's it's obviously a multi-year prep for this to happen, right?
[00:04:13] And I think especially with the turn of events, we talked about this so many times on the show of like, Hey, capital costs something and you can't just burn money and yada yada. Especially if you want to go public. Now, the public wants to see someone actually making a profit like a real company and not burning through cash like crazy.
[00:04:30] You know, they had a pretty tall mountain to climb in front of them. To, you know, change their profile to go public to kind of be, be interesting, you know, make this an interesting public company basically. And one of the major feats, and I think you can fact check me on this, but they were at some point, what, two, two years ago?
[00:04:48] Three years ago. I think they burned through a billion dollars a year in terms of cash per, just like money, literally leaving their bank account never to come back. And then now turn this around. To be what? Sub sub 50 million annual burn or something like this. And on track to go public sorry.
[00:05:05] Go profitable from that?
[00:05:06] Mikkel: We'll probably see when their S one drops. I dunno if it's out yet. Actually I haven't seen it.
[00:05:10] The number I had was from mid 24. Around that time they had like 30 something in burn. So
[00:05:16] obviously it could change if a lot of costs just landed in the second half, who knows? But I doubt that it's gonna materially change relative to what the numbers have been.
[00:05:25] Right.
[00:05:26] Toni: Yeah, so let's just say it like it is. They, you know, did a pretty good, amazing job here going from a billion and burn to something else.
[00:05:36] Mikkel: Burning a series A company a year, basically,
[00:05:38] Toni: yeah, sure, sure, sure the thing though is, you know, and this is the story that there's, they've always been pushing. We are gonna get to kind of, is this BS or not later.
[00:05:49] Um, but one of the predominant stories that they've been pushing is like, Hey, ai, we figured something else internally in terms of ai. I think it's safe to say that a billion dollars in annual savings you know, you, you can't get there with just some tooling replacements. The, the majority of that burn reduction, you know, shows there's some growing revenues in there too.
[00:06:12] The majority of that burn reduction came from people's salaries. Let just, let's just be super clear about this. Right. And, and I think this is also setting this up a little bit. It's like, well, you know, was this AI story of them like replacing, you know, kicking out Salesforce, kicking out workday, you know, doing all, you know, I think some posts were saying.
[00:06:32] We, you know, that they replaced over 1,200 apps from four to 500 vendors with the AI strategy. Was that that helped them to, you know, save a billion dollars a year? Or was it really just the nicer, bigger story that they could put front and center while they were showing all of these people, you know, with the doors?
[00:06:51] And I think. There's probably a little bit of a mix in here. I think we have a pretty good understanding kind of what the lopsided view on this is. But I think this is what we wanted to talk about today.
[00:07:00] Mikkel: Yeah, so really it's funny how the events unfolded, right? We talked about this. The first story I remember hearing was around them firing a bunch of people. And it's,
[00:07:09] I think it's natural for you and I maybe to have seen that on LinkedIn because of second degree connections working there. Honestly, then, then you get exposed to kind of, when a thousand people lose their job or 700 in in support, lose their job.
[00:07:20] It, it creates a lot of noise, honestly. And I think to your point, making the argument that, hey, now they've slashed 1200 apps. That's how they went from 1 billion in burn to like almost zero. Laughable laughable. I gotta admit, like the one thing I remember from business school going to uni and it's just the one probably is, well, the best way to reduce cost is fire people.
[00:07:43] That's
[00:07:43] like, that's the best, fastest, easiest way. It's where most of the cost goes anyway. And it's the same for software companies. Funny enough, it's not like the server bills.
[00:07:51] Toni: No, but let's just, let's just jump into, you know, because you mentioned people, let's maybe jump into this example first. So one of the things that they were able to replace a lot or kind of bring down a lot is their customer response
[00:08:05] Mikkel: Yeah. Yeah.
[00:08:05] Toni: I think this is more like a, like a customer service bot that people have been talking about. Right. And I think, you know, when, when I'm thinking about, okay, how can you reduce the amount of calories you, you pay? I think that's a super straightforward one.
[00:08:19] Like we can, you know, it's unclear how much that really kind of reduced cost, but probably quite a bit. Right. So, and I, and I saw them site that they went from 12 minutes average response time. And they have this B2C side that's kind of taking so much, so much volume here, right. And super simple questions obviously for them.
[00:08:37] But they took it from 12 minutes to two. Right. That's, frankly, that's not, that's not bad, right? Kind of we can poop all over the place for all the other things, but that one is really actually not bad. And then, you know, in connection with that, you also see you probably don't need as many folks there anymore.
[00:08:53] Right? So I think, I think that is the clearest thing for me where I'm like. Okay. You know, we are not, we are not talking about the Salesforce license cost. Let's just say that kind of, that's actually, I think that's bs. And I, and I think Sebastian, what's his last
[00:09:08] Mikkel: W
[00:09:08] Toni: C
[00:09:10] Mosovski. Like he yeah, I don't think he's saying that, you know, not sending money to, to Salesforce is what
[00:09:17] Mikkel: no, no, no,
[00:09:18] Toni: around. Right. But I think the first and clearest one is really on the, Hey, they deployed this customer service thing and kind of helped, you know, things to go better. Right. And I think that's pretty believable. I think that's pretty believable.
[00:09:30] Mikkel: mean supposedly it's take taken care of two thirds of all the requests. Right. I think you also hear similar kind of rumblings from Intercom and the, like, that they're able to resolve a bunch of tickets. Right. So
[00:09:39] I, I think that makes complete sense. It also, I. It is also one of the more cost intensive areas for sure, for that company because there will be challenges, as far as I understand, between merchants and buyers.
[00:09:50] There might be disputes there, which are just more complicated to handle. Right. And yeah, so supposedly that solution is doing the work of 700 folks, which is why they've been cut from the roster apparently. Right. So the, here's, I think the tricky part is making the leap from. Having cut people to AI being the solution.
[00:10:08] Right. And I think you and I have talked on and off air probably about this already, where it's, Hey, is that shaping IPO conversation? I don't know if we're gonna get to that in a second. Probably. We are . So just maybe park that for now.
[00:10:19] Toni: That's right. So, and I think kind of this is, this is one, one piece where it's like extremely clear that there's like cost savings include here.
[00:10:28] I think the topic that comes up most of the time when we're talking about Klarna is apparently the internal chatbot, Kiki. Which, you know, even when you say that, I think it's, it's kind of cute, but also sounds kind of funny.
[00:10:39] Right. And basically they're kind of two parts to this story. One is. What was the issue that they encountered then? Number two, then how did they solve this? Right. And I think the number one part in terms of what's the issue, and you know, Sebastian talks about this in this like couple of podcasts actually kind of quite intensively.
[00:10:57] It's like he's talking about I. You know, in different words, but he's talking about data being dispersed or fragmented or knowledge being fragmented and kind of being all over the place and kind of the sheer amount of cost of employees trying to find that data and tapping into this and, and all that stuff.
[00:11:14] Right. And what I found pretty interesting was apparently they wanted to solve this problem. Sounds like a really. Like a high flying, edgy, you need to be valued $45 billion to even think about solving this problem internally, kind of problem. And and they were looking apparently at Wikipedia and they were looking at other places where knowledge is being kind of
[00:11:32] Mikkel: the CEO ask his blood Boy and, Yeah.
[00:11:34] Toni: Yes, yes. And what they came out of is or come, came out with, but you know, out of that research, they came out of it and kind of had an idea, an epiphany. That they basically wanted to use a knowledge graph. Yeah. So what is a knowledge graph? It's it's, it's a, it's a, you know, a different kind of database.
[00:11:55] It doesn't work in tables, like in columns and rows, like how you and I think about databases. I. It's not called a relational database, by the way, super confusing, but basically kind of, it has think about like a network, like a social network and with nodes and with like lines connecting those nodes and arrows and all that stuff, right?
[00:12:13] Kind of think about this meme of this guy that's like on the wall with those red kind of you know, threads connecting that, but organized, right? That, but organized. That's, that's what a that's what a graph is in general, right? And
[00:12:26] Mikkel: I would've, I would've
[00:12:28] bet you said something with a sheet there and you know, columns and, but okay. Yeah, yeah. Keep going. Keep
[00:12:33] going.
[00:12:34] Toni: What do you mean? What do you mean?
[00:12:35] Mikkel: Like a graph. You're like the numbers guy, the data
[00:12:38] guy, and it's like, Oh, sheet, I wanna move some graphs. It's like, no, no. It's like a
[00:12:41] spider
[00:12:42] Toni: and and it's also fancy, you know, it's like, you know, a knowledge graph and stuff. So, that's what they actually started building, right. And what is pretty interesting there is they you know, obviously sucked up you know, data from slides and from notion and from like other unstructured places, and made this accessible for their employees to ask questions.
[00:13:01] Right. And I think that particular one. And we are gonna, we are gonna discuss, build and buy action in a second, but that particular one, and I would actually say all the examples, which is kind of cool. All of these examples have either since spawned. Industry of, of tools that are kind of solving this or have been doing this, you know, around the time, maybe a little bit before.
[00:13:21] Right. And this particular knowledge graph on like this you know, all of this tech stuff is basically I think we heard about Glean now. They, someone, someone else mentioned Sierra there's like enterprise search with a perplexity, kind of, they're doing all of that stuff basically, right? And the.
[00:13:37] I, and you know, the, the first application that they basically have there was a internal chat app called Kiki. That they're then, you know, used to answer, you know, all kinds of different questions, right?
[00:13:47] Kind of, that's really the starting point. So they're going from, you know, fragmented data to, Hey, here's one place where you can engage with that.
[00:13:55] And that one place is, by the way, slack. It's pretty, pretty straightforward
[00:13:57] Mikkel: I think the cool thing here is also like when you think about the amount of tools, companies tend to just add your data, get dis dispersed in so many different places, which in itself is actually a risk. It's a little bit of a risk. You gotta be, be mindful of, and I think obviously Klarna, they will have some kind of a CDO or CIO who cares about that stuff, where it's like. They're gonna love sitting in a meeting discussing how we can consolidate, cut, you know, cut some tools and just consolidate the data for not just safety purpose purposes, but also efficiency.
[00:14:29] I think the other cool thing with the internal chat search, sorry. With the enterprise search Kiki, it's served 250,000 queries so far, according to Klarna.
[00:14:37] I don't know how, how long ago this was, but it's roughly 2000 daily requests. They're, they're taking through that solution. Think about it. They're like 4,000 people. It's half the
[00:14:46] team running some kind of query every day. So definitely, you know, great internal tool they most likely have built there. And I also can see the appeal. You are like in a finance industry, usually very regulated by the way. Right? Incredibly regulated. So for Glean or someone else to pass all the security checks will be, will be difficult. Like honestly, we've heard stories of. Companies much bigger than glean getting thumbs down in this sector in Europe because of security purposes.
[00:15:14] Like, so, so I think building it yourself will still unlock that that utility.
[00:15:19] Toni: Absolutely.
[00:15:20] I think, I think where the Klarna guys went further than Glean, by the way, is Glean is very much focusing on, you know, your Google drive, your notion, your slack, kind of, that kind of stuff.
[00:15:32] The, the Klarna guys apparently also integrated the EAP like where the finance stuff is. I think they did, you know, similar stuff with, you know, we're gonna get to that, to their Salesforce and kind of their CRM stuff, the HR systems and so forth. Which is really, which is really interesting because you can do a bunch of things on top, right?
[00:15:49] And basically creating that knowledge graph of the entire organization. Like first of all, that's a big undertaking, right? The other thing is you kind of take all of the fragmented data and knowledge and kind of create a place. That you as a human can't navigate either, by the way, but the AI can, which is why this is so interesting, right?
[00:16:06] Which is also, by the way, why we are building the knowledge graph for the go-to market, yet a of right kind of, that's actually kind of where we are seeing, you know, some folks like lean and, and perplexity not spending a whole lot of time and, and trying to fix that is really kind of trying to understand how does the whole go to market actually hang together?
[00:16:22] How does that data and knowledge kind of hang together and then deploying AI in top in order to kind of go through this right. What I think Klarna did extremely interestingly here, and this is now moving from a general written staff knowledge graph to also applying it for tools.
[00:16:38] Is they famously deprecated Salesforce.
[00:16:41] Right. And, and kind of thinking about that, right? Completely insane. Like you don't buy glean or perplexity and then deprecate Salesforce. It doesn't, it doesn't work like that, right? Kind of that, that's not the, the way you can go about this. But basically what they realized is okay, you know, the CRM is also just a it's just a fucking database that people need to maneuver.
[00:17:05] And what they started to do once they, you know, took all of that structure away from the crm, it's like, Hey, you know what, actually, you know, we can now have people use a different interface to get the same stuff done. Right. And I think you and I had a really, you know, you know, prepping this show back and forth kind of.
[00:17:21] Yeah. Did this CM thing really happen? Is it true? I think we looked up how many CRM admins they still have in the organization and oh, their Salesforce certified and did Mark Boff, you know, rebuke this whole thing? So I think there's some truths to the whole thing. Whether or not this was the right move and whether or not this is, you know, blowing this up more than it should, you know, I don't, I don't a hundred percent know yet.
[00:17:44] Right. But what we do know, at least from some of the interviews that we've seen, some of the information online, we've seen. Is that, you know, many workflows that used to happen in this year. I'm in a, like a desktop kind of interface, apparently now happening on Slack. Funny enough, obviously Slack is owned by Salesforce, so kind of, they didn't quite leave Salesforce in that regard, but a lot of those workflows actually kind of have been potentially ported away et cetera.
[00:18:09] Mikkel: Yeah, I mean, I just find it funny that they went from firing a bunch of people and that being extremely negative. Plus they wanted to IPO What do you do when AI all of a sudden is all the craze? Well, you have an opportunity to change the story. I'm sorry. Like, and then hearing the CEO go, oh, I was so embarrassed when Mark Benioff was asked about why Klarna's churn.
[00:18:30] I was like, that's such bs. Come on. Even, even if it's true, you churn from Salesforce. That's an amazing thing to get, you know, get mentioned there ahead of an IPO. I'm sorry. This is like a wonderful story. You have to realize this company is competing with another big BNPL vendor. There's actually two.
[00:18:47] There's square, which now called Block, which then was called X, Y, Z, which I, I dunno, probably is gonna be called something else really soon. And then you have Aian. I. As well, right? You need to basically create an investor story of how you're different from those two other massive vendors to justify a massive billion dollar valuation. And I think this was just super convenient. You, number one, are able to change the conversation away from the people you let go and build a story of how you're, you know, basically a forward thinking company using AI to your advantage to slash cost and everything else. Like I think it's a wonderful story to tell for, for investors. Do we know whether they turned on Salesforce? Like truly? I think probably Mark Sales, mark Benioff would have said their customer if they actually were like, I, I think he would have said that most likely. The thing is though, to your point, we literally checked before recording, they have folks who are CRM specialists on payroll who still working there today who are specialized in Salesforce.
[00:19:44] Doesn't mean that they're operating Salesforce, but probably there's something still running. We're gonna talk about deal in a minute, by the
[00:19:50] way, where. I just find it very convenient that they are able to replace, in that scenario, a bunch of workflows that they don't need the hr software to do, but they will need the HR software to do payroll.
[00:20:03] Right. There are still
[00:20:04] some
[00:20:04] things you will
[00:20:05] Toni: and I think actually that gets to the heart of this whole thing which I think is extremely interesting, which is know, we've built up tens of thousands of SaaS tools. You and I know from a marketing perspective, you focus on the one thing it does, right? Like the one thing, not the other 20,000 things.
[00:20:21] The one thing when you go on Stripe's website, it still, you know, run one payments or, I dunno what they're saying, but like it does, they have like eight or 10 other products by the way that people are buying, but that's what they're focusing on, right? And really when you strip it all down, every single SaaS tool probably only has one single purpose.
[00:20:37] Like one single purpose. And I would say, and this is kind of what's kind of interesting here. Some of those sales tools that have a single purpose can get easily replaced if you have the right data. If you have access to the right data, then kind of that stuff can easily get replaced, right? I think to a degree, this is what happened here with Salesforce.
[00:20:56] The key skills that need to be managed is still applied. That's why you have those CRM admins. I think they probably have a different interface if. If they're not still using Salesforce, by the way, to your point, right.
[00:21:06] Mikkel: Or an alternative.
[00:21:07] Toni: the, the yeah, exactly. Well, yeah. Anyway, but I think the you know, the deal story is kind of interesting.
[00:21:14] So what happened here they famously went out to say like, Hey, so Klarna hey, we called Salesforce and we, you know, planning to kill Workday. You know, in the coming weeks or something like this, right. So, okay. Well that, you know, that was interesting. And then later, you know, came out that, oh, we switched to deal.
[00:21:31] It's like, well wait a minute. That, that feels, you know, I get the Salesforce to Slack, you know? Okay. Yeah. Sure, sure, sure. No, I get it. But the, the, the competitor A to competitor B, that's kind of weird. So we double clicked on this for you guys, so you don't have to, and basically, you know, and I think that puts it into an interesting perspective.
[00:21:50] They basically realized that, you know what, running payroll on like 50 different countries
[00:21:56] Mikkel: Difficult.
[00:21:57] Toni: we're not, you know, we're not gonna do
[00:21:59] Mikkel: Yeah. Taxes alone, like sure, AI can tell us, but do we wanna trust it with that kind of
[00:22:03] information? What if we do wrong taxes for our employees for half
[00:22:07] a year and they get dinged?
[00:22:08] We screw everyone. It's like, no not doing that, not taking that risk.
[00:22:12] Toni: So that core part of the use case.
[00:22:17] They have still with a vendor, what they stripped away from you know, within that switch. You know, I would call it some of the more softer HR things around the organization. So, for example, org structure. Who reports to whom, you know, what's the job description, who does what, you know, those assessments and like some of these pieces basically went into the graph.
[00:22:39] Right. That's, that's how they're kind of doing this. This went into the graph. And then they have core use cases with deal, which is basically payroll and taxes for payroll. That kind of still sit there. Right. And I think also makes sense. He also mentioned another interesting use case around this employee graph piece that they built there, which is basically and again, apparently they're not companies solving that use case by the way.
[00:23:07] So it's basically instead of sending, you know, everyone a Culture Amp or letters you know, questionnaire, kind of tick the box, tick the box, tick the box done which left them a lot of room for subjectivity. Like, you know, how does, it's like, I know Michel Scott read again, but, you know, the team is difficult.
[00:23:24] You know, kind of that kind of thinking. And what they did instead. Using their own tools, using their own graph and rack and what have you. They deployed an AI chat bot to every single of their, I don't know, thousands of employees. And were like, Hey, would you mind spending half an hour talking to this chat bot about, you know, what you think are Klarna strength and benefits and where do we need to improve?
[00:23:47] And so on and so on and so forth. They did this for all of these employees. They spent half an hour on this thing. Then the AI came back with very clear instructions for all the different levels of, of the organization where there are things to improve, right? Sure. You can say, okay, now the subjectivity, you're now just offloaded to an ai.
[00:24:04] But, you know, you know, I've, I've, I've used culture before and I've sat in those painful. Painful employee feedback sessions there. And it was all over the place. And having some help me with that. I am, you know, I can see the value in something like this. Right. And, and again, this is like one of those things where they basically took a use case that comes out of the box with Workday that, you know, then, you know, moved into their own AI and, you know, environment, so to speak.
[00:24:31] Mikkel: I also just feel like it's, we're at a point where the reason a company cons like this and others, by the way, they're not alone considers building their own stuff, is all the infrastructure hasn't been built yet. It's like we discovered gold. Now we need to build up the mining industry and that's gonna take a while.
[00:24:48] Like we need railroads, we need like specialized tooling and all these kind of pieces. It's just not there yet. Honestly, lots of vendors have been quick to add AI to their software, but for most, I think it's also about reinventing the company, which honestly is just way more difficult. And then you look at the disruptions is like years out.
[00:25:08] For me, I think it's a great example, don't get me wrong. I think it's they've built some cool stuff. I, I don't think you can take away from that, from the company. Do I believe this is like a narrative for IPO? Yeah, I totally do. I, I, I have to admit it. Like I, I see. So many problems with just going in and shutting down 1200 tools.
[00:25:26] Like just the amount of work itself it creates is just insurmountable to be honest. Right. And we've not even even discussed the details of ripping out Salesforce. What about all the other dependencies you have, all the other activities you're performing, there's. There's so much there. And, and we've not seen folks on MASK go out and discuss how they've done it in detail.
[00:25:46] We get very little information out of this, even from past employees. So dubious dubious.
[00:25:51] I think what does become interesting to discuss though is given where we are today, I. Should you actually go and build this stuff yourself or should you go and buy it? Because it's not the first time we've had this conversation.
[00:26:03] It's, it's just not the first time companies have considered building their own whatever solution. And I think ultimately what you have to look at is. Do you have solutions out there that can fix your problem? And is it the best use of your resources? E even engineers like cost money, even if it's just time, they're gonna have to spend and you're gonna be left with the maintenance of it.
[00:26:24] Google, they tried to build basically a deal kind of thing way back. They had to abandon it, that didn't work out. And then a buy a solution to take care of that side of their basically their operations, right? So I think. I'm very curious to see how this involves, and also just to hear your thoughts on, on
[00:26:40] this.
[00:26:40] It's still super early. Still super early.
[00:26:42] Toni: so let's talk build versus buy, right? And those are also questions that the Sebastian guy has been getting, you know, asked a lot. And I think he has kind of a, not a strong answer. I think you know, part of his answer is like, well, you know, when we did this thing, these other companies kind of went around or went like as a clear choice that they are now, et cetera, et cetera.
[00:27:04] And you know, hey, you know, we built this use case and you know, half year later we see a company building this use case, right? So there, there's a timing aspect to it, right? You know, and, and the reason why this is weak is because that might be very different for everyone listening. Oh, oh, you didn't build a knowledge graph yourself yet.
[00:27:21] So maybe then that's, you know, that is then the only timing piece here. Right. The other thing you mentioned, which I also don't think is super, super strong. It's powerful, don't get me wrong, but it's not super strong as the, oh, we we had a bunch of learnings in AI doing this stuff, right? Kind of, that we then use for us internally, yada, yada, yada.
[00:27:39] I'm not so sure. I think there's certainly learnings here but does that really outweigh all of this other stuff that kind of, you know, it's very expensive learnings. Let's just put it
[00:27:48] Mikkel: Let's say they just used all the resources to figure out a great AI solutions for their customers instead
[00:27:54] of themself. Like very different. Very different. They would still have
[00:27:57] learnings, by the
[00:27:57] Toni: they, and they, they did by the way. And, and, you know, all of that stuff. But still, you know, it's sad. It, it's a little bit like when, when I hear people say like, oh, we have a, you know, we invest in a massive SCR team so we have a good talent pipeline for the AEs. It's like, like, that should be, you should, it should be a part of the reason.
[00:28:16] But is it the, you know, that that's a very expensive talent attraction kind of, you know, thing. You're running there, so anyway, right. So I'm not sure, is that the right thing? I do believe, I do believe in this whole issue of fragmented knowledge and data. I believe in that I think there is, that there needs to be a way out, unclear if, if this is a gle, if this is a perplexity, if this something like ADA for the market space and clear that's the answer.
[00:28:42] But that certainly is the problem. Like that certainly is the problem. Right. And what I then think is kind of cool, and this brings me to my next kind of thought here in this bill versus bias, is like once you kind of have the data stuff under control, like once you have the underlying kind of data piece under control, whether that's, you know, data from Notion and you know, Google Drive and, and other places or your Salesforce, your Zendesk, you're gong you know, building an interface on top for people to.
[00:29:09] Work with them, manipulate it. It becomes a fairly simple. Thing actually. So, and, and the way kind of we are thinking about this, so kind of, we have a, we have a graph for the go-to market. You know, you can hook it in and right now ask CLO to build a dashboard. Like, okay, oh wow, Tony. Awesome. But you know, the leap isn't so far to ask CLO to build a forecasting tool or to build a lead scoring tool or to build a, you know, whatever tool on top.
[00:29:34] And. This won't be the a hundred percent solution, but it will be the 80% solution for that specific problem. Right. And if you then go further down the road, if you have all the data under control, whether that's go to market or for hr, or for finance, or for, for, for everything. I wonder, I wonder is a tool on top to solve this, just a lovable bolt.
[00:29:58] Query prompt away is, is that where we are kind of working towards where people just, once they have the data piece in place and the AI understands all of that stuff, is it gonna be so, so super fucking difficult to prompt your way to an interface that a human then can interact with? Kind of that's, that's a little bit of my question that I have kind of where this whole thing is going.
[00:30:20] Right. But. Until we have solved that, I think it's, it's a pretty fast stretch to kind of replace all of those SaaS tools.
[00:30:27] Mikkel: I think the jury is still out there. Also because it is early days, like hearing story about some of these solutions exposing API keys, which is not a great thing. We're hearing stories of folks who build their own little SaaS company with zero coding experience and now can't scale the foundation. It breaks bugs everywhere because they don't know infrastructure or you and I have seen this kind of thing. Building AI software is like 20% AI and 80% like engineering. You still need to
[00:30:54] Toni: No, no, no, no. It's, it's the quote is much more funny than that. It's it's 5% AI and a hundred percent
[00:31:01] Mikkel: Oh, yeah, yeah. Okay. Yeah. Strawberry.
[00:31:03] How
[00:31:03] Toni: the, the, so I totally, totally believe that. But also kind of, Hey, you know, let's not fucking forget, like lovable, they have like tens of thousands of happy users. Like I, I
[00:31:12] Mikkel: not
[00:31:13] Toni: built a website.
[00:31:14] It's, it's, it's, it's nuts. It's nuts. And we still need to grapple with what that actually means. But I think there's I think there will be for the s and b in the beginning and later on for something else, but there will be a little bit of like, Hey, let's just prompt this thing together here. Unclear.
[00:31:29] That's gonna develop
[00:31:30] Mikkel: think the interesting thing potentially that's not being discussed that could play out, if you look at the good old car industry, there's one vendor who did vertical integration. That vendor was Tesla, and it gives them such upside in building software
[00:31:46] that the
[00:31:46] others
[00:31:47] Toni: a, what a great
[00:31:48] Mikkel: Because, Because, yeah, great.
[00:31:49] But because the others have to go to all these OEMs, all these suppliers, and have them build certain things in certain
[00:31:56] ways. If you now contrast this to software. Klarna is building basically their own apps internally. This is like not only a mode, it's gonna be hard for the competition to copy because they don't know how they've built it and it's gonna potentially give them an upside in how they wanna expose data, where, how they wanna integrate the stack, what they wanna be able to do. It just comes with all the burdens of maintaining the software and actually building it, right? So I think it's gonna be really interesting to see how this develops, whether someone's gonna make a bet on. Basically saying, no, we're not gonna buy all this good old shelfware, which it is now. We're gonna develop our own stuff because it's much more efficient.
[00:32:29] It's not just about the cost. It's actually about what we, what value we can create as a business. And I think that potentially remains to be seen. It's also gonna be very different. All vendors who sell software, they have an ROI calculator and some ROI quote. And then there's the reality, which is, did you get some RI from that tool?
[00:32:47] It's like, well, don't know, maybe.
[00:32:49] Toni: I, I gotta say, I gotta say kind of this whole Satya Nadela on like, oh, S'S dad, and like, he didn't say those words, but like, something like that. Very. A couple of months ago felt very far away for me, like very far in my head. I was like, oh, come on. Like you're just trying to sell your fucking copilot.
[00:33:09] Come on, let's, let's not blow this out of proportion. I think what has changed in my mind I started, so I didn't use you know, credit where credit is due. I didn't use Lovably used, base 44. I could just sit there as a non-technical guy and kind of, you know, prompt this thing. And it was, it was generating good stuff, like, you know, and I think that is different, you know, for me, you know, if it can generate a website because I'm non-technical, that means it can generate a UI for me.
[00:33:36] It means, you know, I can do all of these things for me. Right. So I think there's some stuff that's changing. just to wrap this up, right, and maybe you get kind of like a a, a last thing here. Do we believe this AI thing from Klarna is BS or not for their IPO only? Or you know what's, what's our verdict here in the end?
[00:33:53] Mikkel: Let's just say, I think they're using it to their advantage. They're probably building a ton with ai. I think they are. I think it landed at an opportune moment for them to say that they're cutting certain tools to get some press. They're shaping a narrative for sure. I don't think one rules out the other.
[00:34:08] I know it's kind of a, you know, a neat way to cover my own ass, I guess. But that's just how I see it. Like I do believe they've built these pieces internally. I do think they must have some other mechanisms in place if they've got it. Salesforce, you and I even talked about those types of contracts, usually multi-year usually.
[00:34:25] So it could be that they are still actually operational but just used much less.
[00:34:30] Toni: Yeah, I think it's a head fake. I think it's a great way to own the narrative when you shed half your workforce, by the way. But I also don't believe that, oh, we want to fire those, you know, thousands of people. Let's build an AI initiative so we can own the narrative. I don't think it happened like this.
[00:34:49] I think, I think they were like, Hey, we need to let go of those people. And then in parallel there was like, Hey, let's be more pro, you know, increase productivity using ai. And suddenly that took off and worked and worked and worked. Then they kind of made a thing out of that suddenly that took off and worked and worked.
[00:35:04] And I think now they just went all in with this. And I think probably all into a degree where it almost kind of, you know, flips over on the other side. It's like, I think, you know, there probably a couple of areas where you shouldn't just try and deploy that stuff, but that's our extremely professional,
[00:35:19] Opinion on this. And, everyone for listening. If you haven't, you know, hit subscribe yet. Please do. And otherwise see you next time. Have a good one and thanks Mickel.
[00:35:27] Mikkel: Bye Toni.