Klarna CEO: SaaS is Dead: Why Systems of Record Will Die in an Agentic World (YouTube Video Transcript)

(00:00:00) Your YouTube transcript will appear here (00:00:00) We've gone from 7,000 people, we're now (00:00:02) below 3,000. We've shrank 50%. And I (00:00:05) didn't ask for a single dime to do all (00:00:07) this. And the reason for that is because (00:00:08) I've seen the acceleration of AI and I (00:00:10) know we can ship all these things on the (00:00:12) existing organization. It's 2030. How (00:00:15) many employees do you have then? 2,000? (00:00:18) >> No. It may very well be even less than (00:00:20) that. (00:00:20) >> No, but listen now. We have an (00:00:22) incredible episode today. Seb from CL is (00:00:24) probably one of the leading figures in (00:00:26) how to implement and use AI effectively (00:00:29) to shrink headcount and make your (00:00:31) business way more efficient. This was (00:00:32) one of the most wide-ranging (00:00:34) conversations we've had. This is what I (00:00:35) signed up for. It is stressful. It was (00:00:37) hard as hell, but this is what I wanted. (00:00:40) The next thing that's going to hit (00:00:41) everyone bad is the switching cost of (00:00:44) data because (00:00:47) ready to go. (00:01:00) >> Sebastian, it is so good to have you in (00:01:01) the studio, dude. We've done this (00:01:03) before, but to have you here in person (00:01:05) is fantastic. So, thank you for joining (00:01:06) me. (00:01:06) >> I am so happy to be here. This is going (00:01:08) to be a lot of fun, (00:01:09) >> dude. This is going to be great. So, (00:01:10) this is also going to be the best show (00:01:11) you've ever done. You've done a lot of (00:01:12) shows, I'm telling you already. But I'm (00:01:14) just freewheeling. I had these brilliant (00:01:16) notes, dude. Where the [ __ ] is value in (00:01:20) a world of anthropic and clawed code (00:01:23) wiping billions of dollars off a stock (00:01:26) market? How should I think about that? (00:01:30) >> Uh, you should think that software cost (00:01:33) of creating software is going down to (00:01:34) zero. That's it. So uh and that means (00:01:38) that like everyone will be able to (00:01:40) generate software at any point of time. (00:01:41) So it is a massive change and uh I was (00:01:45) 100% con you know convicted about this (00:01:47) already when I saw this one or two years (00:01:49) ago. So I've been that's been very very (00:01:50) clear to me. (00:01:51) >> If the cost of software creation is (00:01:53) going down, how do we determine which (00:01:55) businesses have sustaining value versus (00:01:58) which do not? So the the key thing right (00:02:01) now is so far the only thing that's gone (00:02:03) down to or not to zero yet but become (00:02:06) extremely much cheaper is the generation (00:02:08) of software. The next thing that's going (00:02:11) to hit everyone bad is the switching (00:02:15) cost of data because so far what you're (00:02:17) seeing is you have proprietary data (00:02:20) stuck in for example the CRM vendor or (00:02:23) the you know other software as a service (00:02:25) that you're using currently. So you may (00:02:27) replicate and build the same dashboard (00:02:29) or build the same processes in your own (00:02:31) tool but all your data is in there (00:02:33) according to their data model according (00:02:34) to their setup. What's going to happen (00:02:36) is people are going to start solving (00:02:37) that problem. How do I get all of my (00:02:39) data from the existing vendor and move (00:02:41) it to the new vendor with the help of AI (00:02:44) through one click that brings down (00:02:46) switching cost and that's when the real (00:02:49) threat to SAS comes. So we had Anishh (00:02:51) from Andre on the show, one of their (00:02:53) GPS, and he said agents in particular (00:02:55) will dramatically reduce the friction of (00:02:57) switching. (00:02:57) >> Yes. (00:02:58) >> Is that the method of which you're (00:03:00) talking about which will allow for this (00:03:01) migration to happen? (00:03:02) >> Exactly. That's exactly what's it's (00:03:04) going to happen. It's happening already. (00:03:05) >> If that is the case, should ERPs and (00:03:10) Service Now and Salesforces (00:03:12) not be dramatically threatened because (00:03:14) the (00:03:15) >> I think I think the stock market woke up (00:03:17) to that in the last few weeks, right? (00:03:18) The question is just like I mean it's (00:03:20) not like any business is going to (00:03:22) disappear overnight because people tend (00:03:24) to stick they have used these things for (00:03:26) a long period of time they like them (00:03:28) etc. The question is that what multiples (00:03:30) should they trade? And if you look at (00:03:32) historically software could trade at a (00:03:34) price to sales, I'm not going to talk (00:03:35) price to e to earnings because some of (00:03:37) them aren't profitable. So it's not an (00:03:38) easy way to compare. But if you do price (00:03:40) to sales, they've been trading at 20 30 (00:03:42) and now they're down at 510. But if you (00:03:44) look at utilities, normal companies that (00:03:47) are more utility, they may trade at 1 to (00:03:49) two. So I from that perspective, you (00:03:52) would argue there is still some (00:03:53) unfortunate potential to come down even (00:03:56) further. Is it going to be look at CHEG (00:03:58) in the US right they're now trading at (00:04:00) 0.2 Chat GPT was seen as basically (00:04:03) wiping out their business a few years (00:04:05) and now they're trading at at a (00:04:06) depressed value and now the revenue is (00:04:08) also coming down actually 30 40% last (00:04:10) time I checked is that going to happen I (00:04:12) don't think so that's probably too (00:04:14) extreme 0.2 would be very extreme for (00:04:16) some of these companies. But is it not (00:04:18) on like is it likely they could come (00:04:20) down to one or two? Yes, I think so. (00:04:22) >> My question to you is that there's this (00:04:23) kind of consensus from all investors (00:04:26) which is always a worrying thing. Um (00:04:28) that if you think that we're really (00:04:30) going to vibe a lot of these tools (00:04:31) internally, you've never worked in a big (00:04:33) organization. the permissions, the (00:04:36) hierarchy that ensues with the (00:04:38) implementation of these tools, we are (00:04:40) not going to see large companies vibe (00:04:42) code mission critical systems and they (00:04:44) will keep the largest systems of record. (00:04:47) How do you think about that when David (00:04:49) Sat says that? (00:04:50) >> No, I understand that some people are of (00:04:51) that opinion. uh I am not because I (00:04:54) think that like one thing is also like (00:04:57) currently the way AI is set up and I've (00:04:59) already started seeing people doing this (00:05:00) differently but one thing is (00:05:03) right now AI is allowing us to reinvent (00:05:05) the wheel all the time right so if you (00:05:06) come in and say I want to write this (00:05:08) piece of code somebody else prompted the (00:05:09) same AI the exact same thing somewhere (00:05:12) else we're still using tons of server (00:05:15) power to do generate the exact same code (00:05:18) what people's going to start realizing (00:05:19) why why don't I cache these things if (00:05:21) I'm getting the same question if I'm (00:05:23) getting the same or why don't I use (00:05:24) existing open source components and (00:05:26) reuse software like what if if software (00:05:29) becomes more like Lego pieces that you (00:05:30) put together that perfect it's going to (00:05:32) be more and more efficient to just like (00:05:34) bundle things together and this also (00:05:36) means that things like what you're (00:05:37) talking about is like production ready (00:05:39) you know security (00:05:42) assured and all these things they will (00:05:44) become more standardized building blocks (00:05:46) and I'm I think in the future I'm not (00:05:47) sure AI was even going to code that much (00:05:49) just going to pick some pieces together (00:05:50) and stitch them together to to come to (00:05:53) what you really need which also actually (00:05:55) means less uh need for compute. (00:05:58) >> The the other argument and I totally (00:06:00) hear that but the other argument is (00:06:02) enterprise software spends about 8 to (00:06:03) 12% of company budgets and you look at (00:06:05) that and you go well hang on a minute (00:06:07) our core business is X (00:06:09) >> Mhm. (00:06:09) >> why the [ __ ] are we building a Monday (00:06:13) replica or a you name it replica? That's (00:06:16) not our core business. Why spend the (00:06:18) internal resources on it? Yep. How do (00:06:20) you justify that matter? (00:06:22) >> I I think that's to some degree correct. (00:06:24) But like if funny the same weekend that (00:06:27) this whole Claudebot thing exploded on (00:06:30) X. I was actually sitting myself and (00:06:33) playing around with a project which I (00:06:34) was just calling company in a box, (00:06:36) right? And the idea was just I was just (00:06:38) I just wanted to test a little bit. And (00:06:39) the idea was to do do very similar what (00:06:40) Clot did but more on a company like for (00:06:42) a small company and I just put like a (00:06:44) small workspace in there was accounting (00:06:46) and in that I put the an open source (00:06:48) accounting software and then I put like (00:06:50) a CRM and I put an open source CRM and (00:06:52) then I put a claw agent on top of that (00:06:54) and then I told my claw agent hey can (00:06:56) you bookkeep this invoice for me or hey (00:06:58) can you set up this customer account for (00:07:00) me on top of that software right and it (00:07:02) worked really really nice. I just wanted (00:07:04) to test the idea because the point is (00:07:06) that and that's actually where it's also (00:07:09) I see the risks of even more uh jobs (00:07:12) being threatened because to some degree (00:07:14) if I'm a small company today then I may (00:07:16) have an accountant uh firm that's (00:07:19) helping with accounting and those are (00:07:20) the ones I would email like hey can you (00:07:22) fix this invoice or how how much money (00:07:23) do I have on my cash you know what's the (00:07:25) current P&L look like etc. But now I (00:07:27) have Claude as an accountant on top of (00:07:29) the accounting open source software and (00:07:32) then I'm just asking hey bookkeep this (00:07:33) you know bookkeep this invoice or check (00:07:35) me my balance and it works really really (00:07:37) well. So I think I'm not saying I don't (00:07:39) think the plumbing firm the electrician (00:07:41) of the future will code vibe code this (00:07:44) themselves definitely not they will buy (00:07:46) offtheshelf products for this but the (00:07:48) thing is the question is most of our ERP (00:07:51) systems that we see today or software as (00:07:53) a service because coding was so (00:07:55) difficult and hard are still fairly (00:07:57) siloed right they don't they are not (00:08:00) broad in their spectrum what they cover (00:08:02) and the kind of winner of the future is (00:08:04) much more likely to be extremely broad (00:08:06) they're kind of coming with a clawbot (00:08:08) or claw of companies like those (00:08:10) services. That's how how I think the (00:08:12) future of that kind of things is. It's (00:08:14) different for a company like CLA because (00:08:16) in our case to some degree this is the (00:08:19) operating system of the company and what (00:08:23) we realized when we looked at SAS and (00:08:25) all these things already two years ago (00:08:26) which is why we started closing down SAS (00:08:28) for us was because we need to provide (00:08:31) our AI the best context. We need to (00:08:34) provide as good context as possible to (00:08:35) be able to perform a job. And if your (00:08:37) data is separated in these silos, a (00:08:40) little bit in this SAS, a little bit in (00:08:42) that SAS, a little bit here, here's all (00:08:43) the project management stuff, here's all (00:08:45) the product definitions, here is the (00:08:47) accounting stuff, here is this, here's (00:08:49) that. You it's it's just harder to (00:08:52) provide it the appropriate context, (00:08:53) right? So to us it was like no we need (00:08:55) to reimagine the tech stack with AI (00:08:57) first being AI native and incorporate AI (00:09:00) and deterministic and probabistic code (00:09:02) into one tech that becomes the operating (00:09:04) system of the bank and I think that (00:09:06) that's like the future of larger (00:09:08) enterprise and that's why to us we're (00:09:10) very mindful like we do still use some (00:09:12) SAS for sure like we use slack as an (00:09:14) example today like which is a salesforce (00:09:17) company right so that (00:09:18) >> you should use slashwork it's it's one (00:09:20) of our it's a compet (00:09:22) Yeah. Yeah. Happy to try it. (00:09:24) >> We incubated it. (00:09:25) >> Happy to hear that. So, you see what I (00:09:26) mean? So, it's just like for a large (00:09:27) company, I think that like obviously not (00:09:29) everyone needs to reinvent everything. I (00:09:30) don't think the plumbing firm will (00:09:32) reinvent. I don't think they're going to (00:09:33) wipe. That's not the point. Uh I think (00:09:35) they will buy something that looks like (00:09:36) Clawbot or company in a box kind of (00:09:39) thing. Um (00:09:40) >> totally get you. It's the idea of kind (00:09:41) of the compound startup and the benefits (00:09:43) that come from not having to integrate (00:09:44) with 50 different providers. So, totally (00:09:46) get that. Do agents not just make that (00:09:48) easy though? The agents not just make (00:09:50) the data migration between different (00:09:51) tools from thirdparty providers way (00:09:53) easier. And actually we'll be looking at (00:09:55) this going it was a ridiculous idea to (00:09:56) ever think that we needed to own every (00:09:58) part of every element. (00:10:00) >> Yep. That's exactly you're exactly (00:10:03) right. That's what it's going to happen. (00:10:04) >> So but if that's the case, why do you (00:10:06) need to vibe code it all yourself and (00:10:07) build it all yourself if you're going to (00:10:09) have agents that are able to move data (00:10:11) between different products much more (00:10:12) easily? (00:10:13) >> It depends on what kind of company you (00:10:14) are. As I said like if you are a if you (00:10:17) are a plumbing company maybe Claude will (00:10:20) offer a solution for all of this (00:10:22) anthropic or maybe there will be (00:10:23) somebody else who kind of uses claude (00:10:25) and empowers this kind of company in a (00:10:27) box experience that I think is still the (00:10:29) kind of unknown answer. We don't know (00:10:31) what's going to happen. (00:10:32) >> Custom support is one I just cannot get (00:10:33) as a category because there have been 14 (00:10:35) players funded with over 100 million in (00:10:37) the last 15 months. Uh and then you have (00:10:40) all the existing incumbents and then I (00:10:41) speak to Ariel at Navan Jackad Wallix (00:10:45) >> and they're building their own. (00:10:47) >> Are you building your own custom? You (00:10:49) are. (00:10:49) >> I mean we were one of the early ones, (00:10:51) right? So we this is actually one of the (00:10:52) things we were surprised but I think I (00:10:54) announced already in 23 that you know (00:10:57) our AI customer service had you know (00:10:59) done the equivalent of 600 agents jobs (00:11:02) and it it caught a lot of attention at (00:11:04) the point of time. Now, you know, media (00:11:06) always tends to like simplify these (00:11:08) stories a little bit. The truth was that (00:11:09) at that point of time, you know, our (00:11:12) customer service was doing very simple (00:11:13) questions. It's like, hey, did I pay (00:11:14) Clana? Yes, you did. Okay, thank you. (00:11:16) You know, like so obviously that wasn't (00:11:18) that hard to to do. Uh but it's still (00:11:21) like I mean to some degree large company (00:11:23) like ours, what do we do? We try to (00:11:24) improve our product partially. We try to (00:11:26) do that to for fewer people to, you (00:11:29) know, contact us and ask, hey, it isn't (00:11:32) working or I'm not I don't understand (00:11:33) what I'm supposed to do, right? them in (00:11:34) part of it. You just want your product (00:11:35) to be so good that people don't feel (00:11:36) they have to do that. So, we've always (00:11:39) tried to reduce customer service calls, (00:11:41) right? The only to me shocking (00:11:43) experience back then was that like we (00:11:45) rolled this thing out, we rolled a lot (00:11:47) of product improvements out. I've never (00:11:48) rolled a product improvement out that (00:11:50) instantaneously took took away 600, you (00:11:54) know, the equivalent of 600 agents worth (00:11:56) of work. Now, these because we don't (00:11:58) hire these people ourselves, they work (00:11:59) for customer service companies, they (00:12:01) just shifted and started working on. (00:12:03) Fortunately, in that situation, nobody (00:12:04) lost their job, but it was still like an (00:12:06) eye openener for us. Like, wow. And then (00:12:08) the point is what you realize when (00:12:11) you're early on that journey is again (00:12:13) for customer service agents, whether (00:12:15) it's AI or humans for that sake, for (00:12:17) them to be able to answer questions (00:12:19) really well, they need as much context (00:12:21) as possible. Where is that context? It's (00:12:24) in the source code of your software. How (00:12:27) does CLA calculate interest? Well, we (00:12:29) can have a a documentation of that, but (00:12:32) at the truth is in our source code is (00:12:35) somewhere deep in our source code where (00:12:37) that interest calculation is actually (00:12:38) explained, right? So even if a (00:12:40) documentation may be inaccurate so what (00:12:42) you realize when you pursue this is that (00:12:44) like customer service isn't just like (00:12:46) hey I need an agent that answers (00:12:47) questions sooner or later you wanted to (00:12:49) read the source code and explain to the (00:12:51) customer how it works. You want them to (00:12:52) provide as much context as possible to (00:12:54) be able to give the right answers. And (00:12:56) that's when you start realizing that (00:12:58) it's not something you in our case at (00:13:00) least we come to the conclusion we (00:13:01) cannot buy it off the shelf because it (00:13:03) actually becomes part of our tech stack. (00:13:05) >> Will every large technology first (00:13:07) company build their own customer support (00:13:08) system? (00:13:09) >> I'm not sure. I think there will be no I (00:13:11) think obviously I I I believe that for (00:13:14) Clon the right thing was to be early to (00:13:18) you know try to find what we can do with (00:13:21) this technology and where it can bring (00:13:22) us and I think it's it's going to be a (00:13:24) competitive advantage over time to (00:13:26) incumbents that haven't done that but a (00:13:29) lot of incumbents will obviously procure (00:13:31) fantastic AI customer service solutions (00:13:33) in order to try to you know reduce the (00:13:36) gap between what we're doing and they're (00:13:38) doing so you know who knows We'll see (00:13:39) what happens. But in our case, it was (00:13:41) very evident that we needed to do this (00:13:42) ourselves. (00:13:43) >> When you said this, it was a brilliant (00:13:45) headline. It was like Seb from Pler (00:13:47) replacing you hundreds. I can't (00:13:48) remember. I think it was 6 or 700. Uh (00:13:52) replace 700. Um and it was a big fur. (00:13:55) And my question to you was do you think (00:13:57) that did more to harm or to help you? (00:14:00) Because for me as a marketer I actually (00:14:02) thought it helped you because it put you (00:14:04) in a AI first CEO camp that very few (00:14:08) public company CEOs are in. (00:14:10) >> Yes. I I it's but it's a good very valid (00:14:13) question and one of the things we also (00:14:16) realized is that I mean like I obviously (00:14:20) like as as a Polish person I think when (00:14:24) when things are about to change I look (00:14:26) at them very cynically right like I'm (00:14:27) just like okay this is what's happening. (00:14:29) I'm not that kind of person that's like (00:14:31) gonna try to gloom things over. I'm (00:14:32) like, "Okay, this is happening. It's (00:14:33) going to be a big change to the world. (00:14:35) How do I adopt? What do we do the best (00:14:37) out of it?" And I'm sure there's going (00:14:38) to come a lot of positive things, a lot (00:14:39) of negative things. Uh, in this case (00:14:41) also, when we announced this, we (00:14:42) obviously had some people being very (00:14:44) frustrated with us, like, oh, you know, (00:14:46) you're you're laying people off because (00:14:47) of AI. People were angry to some degree (00:14:49) as well (00:14:50) >> and and I I respect that. I understand (00:14:52) why, right? And that's why a few months (00:14:54) later, we tried also to go out and say a (00:14:57) different story. you walked it back. (00:14:59) >> Yeah, we don't think so. I think (00:15:01) Bloomberg changed kind of the headline (00:15:03) and then that got misinterpreted. But (00:15:05) what we were trying to say a little bit (00:15:06) later on is that like to me as we (00:15:08) explore this further, we were like, (00:15:10) well, but to be honest, if AI can do (00:15:13) customer service, it means it's going to (00:15:15) be the cheap customer service. It's (00:15:17) going to be the one that everyone gets (00:15:19) because it's cheap and simple. But as (00:15:22) has always happened historically like (00:15:24) when you know when when people started (00:15:26) in factories making cheap clothing or (00:15:28) cheap furniture we started appreciating (00:15:31) artisan things. (00:15:32) >> We started appreciating an artisan (00:15:34) coffee shop and artisan you know (00:15:36) manufactur or furniture that was done by (00:15:39) artisans. So we said the future of VIP (00:15:41) experience will be the human connection, (00:15:44) the relationship and then we genuinely (00:15:45) believe. So we said we need to transform (00:15:47) our customer service from thinking about (00:15:49) it as like okay yes it's officially just (00:15:53) like good customer service but to some (00:15:54) degree when I kind of challenge it (00:15:56) internally I said look what I've seen (00:15:57) has happened is also too much focus on (00:16:00) cost right there's been too much focus (00:16:02) on that. have to rethink this and make (00:16:04) customer service into this human part of (00:16:06) what Clana is and make sure that we (00:16:07) offer everyone who wants a human (00:16:09) connection. That's going to be what VIP (00:16:11) service looks like in the future. Like, (00:16:13) oh, I'm not dealing only with machines. (00:16:14) I'm dealing with a human. You know, (00:16:16) that's what we think at least. So, (00:16:17) that's the message we try to get out. (00:16:19) >> I completely get you, Seth, but I'm (00:16:21) sorry, dude, for being so blunt. That (00:16:23) sounds a bit Silicon Valley idealistic. (00:16:25) And what I mean by that is most people (00:16:27) are not actually as good as we think (00:16:29) they will be. Often customer support is (00:16:31) roles where you're in there for a year (00:16:32) or two and you move on. Yeah. (00:16:34) >> And what you need is Seb, I know that (00:16:36) you have, you know, x number of kids and (00:16:38) you often like to travel here. (00:16:39) >> I thought about these three restaurants (00:16:40) for your romantic Valentine's Day trip (00:16:42) with your wife. (00:16:43) >> That's a really thoughtful agent or (00:16:45) person. (00:16:46) >> The trouble is most people are not that (00:16:48) and it takes training and development to (00:16:50) get there. the lowkilled labor allow (00:16:52) that all social media marketing content (00:16:55) creation that's low level (00:16:57) >> is going to be eroded faster than ever (00:16:59) before. No, (00:17:00) >> but you're right and that's exactly why (00:17:02) we started changing. So one thing that (00:17:03) we've done that has worked fantastic for (00:17:05) us and we're rolling this out. We (00:17:06) started saying okay to your point if we (00:17:09) looked at like these agents that we were (00:17:11) hiring previously even through uh other (00:17:14) companies we didn't really have a (00:17:16) relationship with them. We don't know (00:17:17) who these people were. like some of them (00:17:18) were great but it was like to your point (00:17:20) a mix right and we said how do we create (00:17:22) something very different and so what we (00:17:24) started since then which has worked (00:17:25) amazingly well and we were kind of just (00:17:27) starting to romp it up is we actually (00:17:29) built like our own Uber model which (00:17:32) means that we encourage today we recruit (00:17:35) our own customers the most passionate (00:17:36) customers that live in rural areas and (00:17:39) so forth and we say hey do you want to (00:17:40) work extra or do you want to work (00:17:42) part-time in our customer service and (00:17:44) they actually just like somebody can go (00:17:45) and drive an Uber for a while they can (00:17:47) actually jump on and work for Clana's (00:17:49) customer service. And these are our most (00:17:51) passionate customer. They love our (00:17:53) product. They love how it works. They (00:17:54) know Clana in and out. And now they earn (00:17:57) extra money by actually working on our (00:17:58) customer service. And obviously the MPS (00:18:00) and the customer satisfaction of those (00:18:02) interactions with our customers is (00:18:04) through the roof. So, so we to your (00:18:06) point we needed to change not and that (00:18:07) was what we were trying but it's like (00:18:09) very hard in the Bloomberg article like (00:18:10) I'm not going to blame the journalist (00:18:12) like I think actually if you read (00:18:13) article the original article it's (00:18:14) actually fairly balanced and it kind of (00:18:16) tries to describe this but then the (00:18:17) headline is like they're rolling AI back (00:18:20) you know and then the whole media circus (00:18:21) go on like oh clown has just announced (00:18:23) that they're rolling it back it's like (00:18:24) no that's not at all (00:18:25) >> the truth is no one reads the article (00:18:27) anymore they just read the headline CEO (00:18:29) predicts end of finance (00:18:31) >> I was just laughing with them because (00:18:32) like yesterday on X somebody wrote like (00:18:34) Cla like obvious Cla like company is (00:18:39) going to offer buy now pay later on (00:18:40) rent. It was Cla like it was actually (00:18:42) another company but people didn't see (00:18:43) that. They just say Cla is going to (00:18:44) offer buy now pay later on rent and then (00:18:46) suddenly we're all over the press. (00:18:47) People are calling us like and it's like (00:18:49) what are we going to do? Like it's just (00:18:50) not true. (00:18:51) >> I remember when I was called in an (00:18:52) article former teenager. (00:18:53) >> Yeah. (00:18:54) >> I was like I think this is a broad (00:18:55) spectrum of people right now. Um former (00:18:58) teenager me and Warren Buffett both have (00:18:59) that in common but I'll I'll take it. Um (00:19:01) I sat down with the team the other day. (00:19:02) you know, our job is to find value and (00:19:04) invest in amazing companies. I was like, (00:19:06) if they sell per seat, we're not doing (00:19:07) it. We need to replace jobs like labor (00:19:10) displacement is what we're investing (00:19:11) into and that's the only way that we can (00:19:14) actually return the amount of money that (00:19:16) we need to now to make funds worth it. (00:19:19) >> Yeah. (00:19:19) >> Do you think that's fair? (00:19:21) >> Well, I think unfortunately I think that (00:19:23) is going to happen. Like I don't think (00:19:25) it's I on and I feel like Daario maybe (00:19:28) is one of few that actually is willing (00:19:31) to stick up and say that officially. I (00:19:34) feel a lot of the other executives of (00:19:36) these big tech companies are getting (00:19:37) nervous. They don't want they're seeing (00:19:39) the negative backlash of talking about (00:19:41) this and then they're trying to portray (00:19:42) something different. But I'm I I don't (00:19:45) want to be one of them. I'm more in (00:19:46) Dario's camp. I want to be honest about (00:19:48) the fact that I do think there's going (00:19:50) to be a very big shift. Now in addition (00:19:52) to that then like I think more like Elon (00:19:55) that it might lead to a golden age of (00:19:57) humanity where you know AI does more of (00:19:59) jobs and more people can enjoy (00:20:01) themselves and do other things that we (00:20:03) can have a richer society. That is not (00:20:05) an unthinkable outcome. It could be a (00:20:07) positive outcome. It could happen and I (00:20:09) think it's not unlikely that it could (00:20:10) happen. So I still I'm an optimist at (00:20:12) heart but I also want to be realist (00:20:14) around what's going to happen in the (00:20:16) shorter term and it's going to be a lot (00:20:18) of turmoil in this. Can I ask when when (00:20:20) you said the statement about replacing (00:20:21) 700 customer service agents at Clana, (00:20:24) >> what did you not know then that you wish (00:20:26) you had known knowing all that you know (00:20:28) now? (00:20:30) >> Nothing. (00:20:31) >> H (00:20:33) you haven't had anything that you've (00:20:35) seen that has changed your mind? You you (00:20:38) know what Harry like in 2015 in 2015 (00:20:41) >> I sat down with my management team at (00:20:43) that point of time we have been trying (00:20:44) to compete with Stripe and Aden for five (00:20:46) years and we were like losing it like (00:20:49) they were just you know crushing us and (00:20:51) we were like this is it and and and Aden (00:20:53) had just signed with Daniel Xify my (00:20:55) neighbor and I was like this is over man (00:20:57) like we're not going to win in this (00:20:58) checkout wars like payments wars forget (00:21:01) it so we're like what are we going to do (00:21:02) as a company what what are we going to (00:21:04) do and then we sat down as a team and (00:21:06) we're like what's the future of banking, (00:21:08) right? So, the future of banking is (00:21:10) going to be some kind of digital (00:21:12) financial assistant wakes you up in the (00:21:14) morning, say, I I checked your mortgage, (00:21:16) you're overpaying like hell, and I have (00:21:18) renegotiated for you, and I can do all (00:21:20) the paperwork. You just need to say yes (00:21:22) to save 50 quid, right? And like that's (00:21:25) the future of retail banking. And that (00:21:27) we said in 2015, okay, so what does that (00:21:30) mean for Clana? We're like, hey, let's (00:21:32) become your digital financial assistant. (00:21:34) let's be that assistant that saves you (00:21:36) that time and money and ever since that (00:21:38) it's been like where are we going so all (00:21:40) these things now obviously I didn't (00:21:41) predict chat GBT I didn't predict all (00:21:43) this stuff happening but that to me that (00:21:45) we're going to become a digital (00:21:46) financial system and we need the kind of (00:21:48) technology that AI is to uh uh to (00:21:51) accomplish that was you know crystal (00:21:53) clear for me for the last 10 years and I (00:21:55) just been running down that path (00:21:57) continuously and to me it's like (00:21:59) self-driving cars we all know it's going (00:22:00) to happen and first it was a huge hype (00:22:02) and every day we're reading the paper (00:22:03) like oh my god It's happening tomorrow. (00:22:04) Tomorrow everyone's investing and it's (00:22:06) crazy and then the hype dies away. But (00:22:08) like I was always saying people asking (00:22:09) me like what do you think about stuff? I (00:22:10) was like look it's going to happen. My (00:22:12) daughter is not going to get a driver's (00:22:13) license. (00:22:13) >> Do you think you're best positioned to (00:22:14) do that? Like if you look at like the (00:22:16) entry point and where you sit in the (00:22:18) stack are Revolute not in a more (00:22:20) strategic better position to be that (00:22:21) digital financial assistant than you. (00:22:23) >> Uh that's I mean I love Revolute. I (00:22:26) think Nick is a fantastic guy. I think (00:22:27) it's an amazing company. (00:22:28) >> Nick's the greatest CEO I've ever (00:22:30) interviewed. I'm also terrified he's (00:22:31) going to kill me. (00:22:33) I love Nick. He's amazing. I've always (00:22:34) felt very competitive with Nick and you (00:22:36) know and I think he feels competitive (00:22:38) with me. Uh we have a common friend who (00:22:41) Oleg who we talk to a lot about these (00:22:43) things which is funny. Uh but the point (00:22:45) is that um if you look at it Cla has 110 (00:22:49) million customers worldwide. Revolute (00:22:51) now has 65. Right? So I'm twice the size (00:22:54) in number of customers. The engagement (00:22:56) that I have is not as high yet as it is (00:22:59) with Revolute. So people may use us more (00:23:01) less frequently and for other things, (00:23:03) right? So what I'm doing right now, what (00:23:05) we're doing at CLA is moving from being (00:23:07) kind of your infrequent payments (00:23:09) solution to being your high engagement (00:23:12) banking uh provider, right? And that (00:23:14) that transition is going extremely well. (00:23:16) It's going extremely well and it's (00:23:17) accelerating. Uh people are adopting our (00:23:20) banking services at a very rapid pace. (00:23:22) >> Do you look at Robin Hood? I've had Vlad (00:23:24) on the show multiple times and I think (00:23:26) Vlad is astonishing. He's got 11 lines (00:23:27) of business that produce over 100 (00:23:29) million in revenue. Um, and I think it's (00:23:32) an incredible model for moving from (00:23:34) entry product BNPL or for him you kind (00:23:38) of frequency trading um, into the (00:23:40) fullstack banking provider. Do you look (00:23:42) at him and do you take lessons from (00:23:43) that? (00:23:44) >> Yeah, absolutely. But we all have (00:23:45) different entry points. I mean, Revolute (00:23:47) to some degree was like an early like, (00:23:49) hey, I'm an expat. I travel a lot in (00:23:51) Europe and like this is better for my (00:23:53) currency or I trade in crypto. That was (00:23:54) kind of the early adopter thing. Uh I (00:23:57) think you know uh Robin Hood was like (00:23:59) I'm trading I love trading. Cla is very (00:24:02) different like our customer is like I (00:24:04) shop online I do shopping you know we we (00:24:07) skew more female than male. We have a (00:24:09) very different brand. I would think of (00:24:11) us more like a lifestyle brand like a (00:24:12) digital version of American Express. (00:24:14) That's how I think about Clana. I think (00:24:16) all of these growing fintexs, you know, (00:24:18) I know Davidid from New Bank since early (00:24:20) days, right? Like have a funny story (00:24:22) where like I was down in Brazil meeting (00:24:23) David when he was just leaving Sequoia (00:24:25) starting and and you know, we had a (00:24:27) fantastic conversation. We talked about (00:24:28) the future of of of banking and then he (00:24:30) sends me an email which was like, "Hey, (00:24:33) Sebastian, would you like to like advise (00:24:35) me a little bit? I'm starting this new (00:24:36) company." And I was like, "I'm sorry, I (00:24:38) don't have the time." So, I lost out on (00:24:41) a lot of a lot. I would have been a (00:24:42) great angel investor there. But that's (00:24:44) how life goes. Yeah. You know, that's (00:24:47) how life goes. So anyway, so that (00:24:49) >> I'm going to be honest. I'm not crying (00:24:50) for you. (00:24:51) >> No. Okay. Okay. Okay. Yeah. I don't (00:24:52) think anyone will cry for me over that. (00:24:53) I will cry over myself. That's okay. I (00:24:55) can I can I can fall asleep crying over (00:24:57) that. But but um No, but the funny thing (00:24:59) is that like so you have these fintex (00:25:01) now, right, who are actually starting to (00:25:03) become really big. I mean, Revolute is (00:25:05) really big, Clonet's really big, New (00:25:07) Bank, you know, etc. So, so and but (00:25:10) we're all coming from slightly different (00:25:12) angles. We're all coming from slightly (00:25:13) different. And now the people who are (00:25:16) going to be threatened by this is the (00:25:17) incumbents primarily. Like of course I'm (00:25:19) going to partly compete with compete (00:25:21) with Revolute. I'm partly doing that (00:25:23) already. Um but if you look at like like (00:25:25) they're very big in Romania. Romania is (00:25:27) not a big market for me, right? Like I (00:25:28) think there's the second largest market (00:25:29) or something for Revolute. If I look at (00:25:31) like number of users like Romania is not (00:25:33) a big market for me, you know? I'm very (00:25:34) big in other markets. So like it's it's (00:25:36) it's the uh so that's going to differ (00:25:38) right but the who are whose market share (00:25:41) are we eating you know Barclays you know (00:25:45) uh the Wells Fargo the Capital One those (00:25:48) are the companies that we are going (00:25:49) after right so I don't really see (00:25:51) there's a big conflict between us it's (00:25:52) more the incumbents that are going to (00:25:54) lose customers (00:25:55) >> you spoke about New Bank and and David I (00:25:57) love David [ __ ] legend um but anyway (00:26:00) obviously they got a bank license in the (00:26:02) US and they're very much aggressively (00:26:03) planning the (00:26:05) Is the US the main goal for you? Yes. So (00:26:08) going back (00:26:10) sitting in London, I'm just I feel so (00:26:13) special in you. (00:26:14) >> No, but listen, it's like this. It's it (00:26:16) comes back to what I said in 15. We're (00:26:18) like the future of financial services is (00:26:20) going to be this digital financial (00:26:21) system. Okay. So then the next question (00:26:23) is who is going to participate in that (00:26:26) challenge and why would Clona stand a (00:26:28) chance? Those are the two questions we (00:26:30) need to answer, right? And we say well (00:26:32) first and foremost there's going to be (00:26:34) three types of companies participating (00:26:35) the tech companies Google Amazon Apple (00:26:38) there's going to be fintech revolute and (00:26:40) so we didn't know I mean that revol (00:26:42) hadn't even started but we knew they (00:26:43) were going to come entrance and then (00:26:45) banks right that was kind of our our (00:26:47) view okay so why would clon so like (00:26:49) what's going to be critical for us to (00:26:51) win in that big transformation global we (00:26:55) first if we're just big at that point of (00:26:56) time we weren't even in the UK yet like (00:26:58) if we're only in the Nordics and Germany (00:26:59) we're not going to have the scale scale (00:27:01) to be able to win this big (00:27:02) transformation that's coming. So, we (00:27:04) need to be global and global means US. (00:27:06) Like, if you're not in the US, if you're (00:27:08) not big there, you're just not going to (00:27:09) be big enough and the risk is you're (00:27:11) going to get acquired by somebody in the (00:27:12) US. So, to us, like nailing us was like (00:27:16) super high priority. Super high (00:27:17) priority. Second, we realized that data, (00:27:21) the more I understand about you as a (00:27:23) customer, the more likely I am to give (00:27:25) you that advice that you said, hey, you (00:27:27) know, you should pick up the flowers (00:27:28) when you go there or whatever. So the (00:27:30) key thing we saw is a difference than (00:27:32) Revolute and all the others. We have our (00:27:34) own payments network just like AMX. We (00:27:36) have our own rails. Every time you shop (00:27:38) with CLA, the information that flows on (00:27:40) the rails isn't just the amount that you (00:27:42) purchase for. It's the exact products. (00:27:44) We have the full digital receipt. So we (00:27:47) know you shopped at Sephora, but also (00:27:48) what cosmetics you bought at Sephora. (00:27:50) And the benefit of that is if I'm then (00:27:52) supposed to advise you on your purchases (00:27:54) or your day-to-day finances, I just have (00:27:56) much more richer information that allows (00:27:58) me to provide you good advice. Oh, these (00:28:01) contact lenses were really expensive, (00:28:02) you can get them cheaper, etc. Right? (00:28:05) So, if I want to help people in their (00:28:06) everyday spending like I have more (00:28:09) information. So, we said understanding (00:28:10) the customer at depth is very very (00:28:12) critical and data was going to be very (00:28:13) very important. And so, it was being (00:28:15) global and having a good understanding (00:28:17) of our customer and having a lot of (00:28:18) trust and brand. The other thing was (00:28:20) brand. Build a brand that people relate (00:28:22) to, that people feel emotionally (00:28:24) connected to and not just like a (00:28:26) utility, right? Create something (00:28:27) different. (00:28:28) >> I'm sorry. I struggle and you know, I'm (00:28:29) a proud European. I struggle when I look (00:28:31) at the European uh neo banks and I look (00:28:33) at Chime and I look at Dave and I look (00:28:35) at all the other current and (00:28:37) >> I mean they're all a fart compared to (00:28:39) the market cap of Revolute right now. um (00:28:42) or whatever the private latest valuation (00:28:43) with the market cap is a very different (00:28:44) thing as you know as a public company (00:28:46) CEO like (00:28:48) why if the US is such a focus has it (00:28:50) been such a lackluster performance from (00:28:52) their domestic participants? Well, (00:28:54) partially because competitive (00:28:56) competition is better. If you take your (00:28:58) AMX app in the US and compare it to one (00:29:01) here, it is significantly better there, (00:29:03) right? Your JP Morgan app is (00:29:04) significantly better. So, the financial (00:29:06) institutions in the US are just better. (00:29:09) There's less. And the problem is that (00:29:11) then people try to find an entry point (00:29:13) that is probably, for example, very big (00:29:15) lending and then they, you know, walk (00:29:17) into a lot of subprime and they make (00:29:19) huge losses and or or there other (00:29:21) things, right? So, people struggle to (00:29:22) find that entry point. But we have 30 (00:29:24) million users in the US, right? Almost (00:29:26) like I think it's 28 or something. So (00:29:28) like but but soon it's going to be 30. (00:29:30) Our card is growing at a very rapid pace (00:29:32) in the US. (00:29:33) >> So your core focus is can I turn those (00:29:34) 30 million users from BNPL into core (00:29:37) customer account? (00:29:38) >> Yes. And we saw we launched a card in (00:29:40) the US and I have to be careful now (00:29:42) because we haven't released the new (00:29:43) earnings yet coming next week. So I have (00:29:44) to use the Q3 numbers but if I remember (00:29:46) correctly we're like at two three (00:29:48) million active card holders in just you (00:29:51) know a few months in the US. So we are (00:29:53) transitioning these buy now pay data (00:29:55) customers into full banking relationship (00:29:57) customers at a very very high pace. (00:29:59) >> Totally get when you look at new bank (00:30:01) and revolute moving both aggressively (00:30:03) into the US if you were to put money on (00:30:05) who's going to do better. Who would do (00:30:07) better? (00:30:09) >> That's a good question. (00:30:11) >> I'm a good interviewer. (00:30:12) >> Yeah, exactly. Who is going to do (00:30:14) better? Cla's going to do better. That's (00:30:16) my answer. I I can't tell you between (00:30:18) those two. It's very very interesting. I (00:30:20) think that the uh Well, we'll see. (00:30:25) I I think I have to go with David. (00:30:27) >> Wow. Why? (00:30:29) >> Because I think uh Nick's challenge (00:30:32) right now might be that he is so (00:30:36) distributed. I mean, he's trying to go (00:30:37) for the whole world, right? I mean, he's (00:30:39) launching in Dubai, he's launching in (00:30:41) India, he's launching everywhere. You (00:30:43) know, banking is not your standard tech (00:30:46) company. It's difficult. And so I just (00:30:49) wonder if he's like running out of (00:30:51) bandwidth, right? Like I think it's more (00:30:52) risky. David at least has slightly more (00:30:55) focus. He has a very solid base in (00:30:56) Brazil making a lot of money from there. (00:30:58) He has his Mexican thing and the other (00:30:59) thing that are growing pretty good, but (00:31:01) he then does US that's just like one (00:31:03) more big market. It's very different. (00:31:05) Nick is so thinly spread right now. (00:31:07) >> Question for you, which is you mentioned (00:31:09) Stripe earlier and the competition (00:31:10) there. Um you're a public company CEO (00:31:14) now. I'm gonna I'm gonna go there and (00:31:16) you can do a no comment. I have so many (00:31:18) public company CEOs on the show. I've (00:31:19) never met a happy one. Never. Um it's (00:31:21) like (00:31:23) are you happy as a public company CEO? (00:31:26) Happy is a strong word. (00:31:29) No, but I think I think look to us it's (00:31:31) like we have over the years so many (00:31:33) shareholders, so many employees. Like at (00:31:36) some point of time it's actually easier (00:31:38) to be public for us. We were already (00:31:40) reporting on an earnings, you know, on a (00:31:41) quarterly basis. We're a bank. So like I (00:31:44) don't I'm not sure that the difference (00:31:45) is that big uh to be honest from our (00:31:47) perspective. Um so it is what it is but (00:31:50) like yeah of course if I could own the (00:31:52) company 100% and be private I would (00:31:54) prefer it right but that's not reality. (00:31:57) >> Do and Patrick with Stripe have (00:31:59) advantages that you don't have as (00:32:02) private companies in their ability to (00:32:05) invest in long-term invest in R&D not (00:32:08) think about next quarter quite so (00:32:09) religiously. historically yes but this (00:32:13) has changed due to AI right so the thing (00:32:15) is that I I I mean one of the things uh (00:32:19) I remember vividly is that when we came (00:32:22) to the board and said that like the same (00:32:24) plan that I said from 15 right grow (00:32:27) global get customers across the board in (00:32:29) all countries and then once you have the (00:32:31) love and affection of these customers (00:32:33) and the relationship with them for (00:32:34) day-to-day purchases then basically you (00:32:37) know go deeper with them offer them more (00:32:39) banking like services and increase the (00:32:41) revenue per customer. Right? So that was (00:32:44) it. But we we had this interesting board (00:32:46) meeting where we're like, "Okay, now (00:32:48) it's really going to happen, guys. We (00:32:49) are now going to truly start focusing on (00:32:52) this transition from just like single (00:32:55) payments customer to banking. (00:32:58) And we're going to launch peer-to-peer. (00:33:00) And we're going to launch, you know, (00:33:02) potentially, which is also leaked to the (00:33:03) press already, trading. And we're going (00:33:05) to do like a lot of like the other (00:33:06) banking services that we at that point (00:33:08) of time didn't have. and we're going to, (00:33:10) you know, increase the card and we're (00:33:11) going to increase balances and deposits (00:33:13) and we're going to do, you know, um, (00:33:15) international remittances and all these (00:33:17) things (00:33:18) and and then the board looks at this (00:33:20) like, yeah, you know, these kind of (00:33:21) transitions can be hard. It changes what (00:33:22) the bank is. We had this discussion for (00:33:25) an additive thought. Thank you, board. (00:33:26) >> Yeah, (00:33:27) >> this can be hard. Well, no [ __ ] (00:33:29) >> Yeah, exactly. (00:33:30) >> But, but the funny thing and then they (00:33:32) say yes and I just like I went out and I (00:33:34) was like, why was that so easy? Like I (00:33:36) was like, how come that went so easy (00:33:38) still? like like it makes oh yeah now I (00:33:41) know why because usually when a CEO come (00:33:44) to a board and say we're going to do all (00:33:46) of these new services we're going to (00:33:47) launch all of this new stuff hence I (00:33:50) need to increase my investment and my (00:33:52) cost by $und00 million if it's the size (00:33:55) of our company to be able to do all of (00:33:57) this but I at the same point of time (00:33:58) were showing them a budget I mean cloner (00:34:00) has been shrinking we used to be 6,000 (00:34:03) or over 7,000 6,000 people and we're now (00:34:06) less than 3,000 and I didn't ask for a (00:34:08) single dime to do all this, right? And (00:34:12) the reason for that is because I've seen (00:34:13) the acceleration of AI and I know we can (00:34:15) ship all these things on the existing (00:34:17) organization. We've gone from 7,000 (00:34:19) people, we're now below 3,000. We've (00:34:21) shrank 50%. And the majority of that is (00:34:23) just through normal attrition. We (00:34:25) initially when we had 2020, we did a (00:34:27) little bit of layoffs, but that was, you (00:34:29) know, not that big of a number compared (00:34:31) to what has happened since then. And so, (00:34:34) so that was the reason it was reason was (00:34:36) easy for the board to take decisions. I (00:34:37) didn't ask for a single dime in (00:34:39) investments. (00:34:40) >> It's 2030. How many employees do you (00:34:42) have then? (00:34:44) >> Less (00:34:46) for sure. (00:34:47) >> 2,000. (00:34:48) >> No. I think it it may very well be even (00:34:51) less than that. (00:34:52) >> No. (00:34:52) >> Yes. But it's a But listen again, (00:34:54) relationship cannot be replaced. One of (00:34:56) the (00:34:56) >> We're going to do a show in 2035 and (00:34:58) Sab's going to be the only (00:35:02) >> Yep. Um there was um I think that the (00:35:05) the the thing is that the one thing (00:35:08) that's important for us is (00:35:09) relationships. We have relationship with (00:35:10) merchants as an example retailers and we (00:35:13) have those relationship locally. So I (00:35:14) have people in Portland talking to Nike. (00:35:16) I have people in in China talking to (00:35:18) Sheen. I have people in you know in (00:35:20) Amsterdam talking to Aden etc etc. So we (00:35:23) have over 50 locations where there's (00:35:25) people AI is not going to move those (00:35:27) jobs like that. We need is very (00:35:28) important the relationships with our (00:35:30) partners and the same customer service. (00:35:32) this gonna be I'm still gonna argue that (00:35:33) it's going to be vital to offer a human (00:35:35) connection there. Yeah. (00:35:36) >> So those jobs will remain but for the (00:35:38) rest it's going to be definitely (00:35:40) smaller. So we're shrinking (00:35:42) through natural attrition with about 20% (00:35:45) per year. It's just people leaving they (00:35:47) stay about 5 years and then they move on (00:35:48) which is natural and then what we have (00:35:50) said very clearly is that like we're not (00:35:52) going to recruit. So we're recruiting a (00:35:54) little bit. People then came again on X (00:35:55) like oh it's not true. Look they're (00:35:57) recruiting. It's like come on guys. Yes, (00:35:59) occasionally we hire somebody here and (00:36:01) there, but if you look at the neck, you (00:36:02) go to LinkedIn and look at the insights, (00:36:04) you're going to see how the company is (00:36:05) shrinking. So the point is that we've (00:36:07) shrank 50%. But we also promised our (00:36:09) employees, which is very important. We (00:36:10) said, "Guys, this going to mean we're (00:36:12) going to do much more with much less (00:36:14) people. This is going to make more (00:36:15) profit for us and you're going to share (00:36:17) in that profit." So our employee (00:36:20) compensation has grown almost 50% per (00:36:23) head during the time. So we have given a (00:36:26) lot of that money back and that creates (00:36:28) safety for our uh for employees. They (00:36:30) know that like through this AI (00:36:32) transformation using these utilities (00:36:33) they are you know getting some of the (00:36:35) benefit of that. (00:36:35) >> Weird question. How do you think about (00:36:37) the current state of SBC stockbased (00:36:39) compensation for anyone which is (00:36:40) obviously how we dividend or give stocks (00:36:42) to employees. You you see aggressive (00:36:44) from open AI argument being why does Sam (00:36:46) give a [ __ ] if he doesn't have any stock (00:36:47) anyway? Dilution doesn't matter to him. (00:36:49) Evan Spiegel is the the godfather of (00:36:51) SBC. Um (00:36:55) astonishingly high amounts. (00:36:57) >> How do you feel about the state of SBC? (00:37:00) >> Well, I think it's very clear when we (00:37:02) compare American companies to European (00:37:04) companies, there's a huge difference. I (00:37:06) mean, I think that like American (00:37:07) companies are 5 to 10x more than (00:37:10) European companies do. Uh, and cloners (00:37:13) by that from that perspective, we're a (00:37:15) European company. We have come from very (00:37:16) low levels. we have increased because we (00:37:18) need to stay competitive uh for talent (00:37:21) and talent today can move between US and (00:37:23) EU pretty easily with the help of (00:37:25) companies. So uh but I still think also (00:37:29) like the the thing that's going to (00:37:30) happen there's been these industries (00:37:32) they're called tech and they're called (00:37:34) Finn right financial services they have (00:37:37) all had this amazing thing which is that (00:37:40) you create this service and there's a (00:37:42) huge switching cost so your customers (00:37:44) can't really switch that easily and (00:37:46) hence you create this money printing (00:37:48) machine and then life is sweet right and (00:37:51) then you build these campuses and you (00:37:53) play volleyball in your office and you (00:37:55) go and get free lunches and you know and (00:37:57) you live off the spoils of this money (00:37:59) printing machine in your basement and (00:38:01) that's not how normal business work. If (00:38:03) you're in retail, if you run a (00:38:05) restaurant, you freaking wake up every (00:38:07) morning and you ask yourself, how do I (00:38:09) put the right product in front of the (00:38:10) right customer, bring them into my store (00:38:12) so that I can actually sell them and (00:38:14) make them happy and so forth. You have (00:38:16) to wake up every day and do care about (00:38:18) that, right? And so the point is that (00:38:20) like it this is going to be a brutal (00:38:22) awakening for Finn and tech that like (00:38:25) that's what's going to happen to all of (00:38:26) us like we're going to have to wake up (00:38:28) every morning make sure that we serve (00:38:30) our customers we work really hard and (00:38:32) effortlessly to make them happy with (00:38:33) what we're offering them and it's not (00:38:35) going to be what it used to be right and (00:38:37) so I think that like sharebased (00:38:38) compensation as an example like (00:38:40) everything that was was it really (00:38:42) because you know it made sense or to (00:38:45) what degree was it just a sports and and (00:38:47) I think to some degree it's sports (00:38:48) That's right. In some degree it's (00:38:50) relevant like because some people can (00:38:52) make a huge difference but there's a (00:38:53) balance between the two. (00:38:55) >> There is an incredible number of (00:38:56) incredible CEOs attacking the incumbent (00:38:59) banks. What happens to BNPing (00:39:04) 2035 (00:39:06) >> I think. So that again comes back to (00:39:09) what we talked in 15. Our conclusion was (00:39:11) there's going to be different. So what I (00:39:13) thought was interesting Solomon also (00:39:15) recognized (00:39:17) uh this at Goldman Sachs. So he created (00:39:19) Marcus and then talking about public (00:39:22) companies the challenge was that when (00:39:24) fintech was like everything was high (00:39:26) 2021 you know valuations up Marcus is (00:39:29) the best thing a Goldman ever did and (00:39:31) then suddenly what ends up happening is (00:39:33) you know Marcus turn and then it's very (00:39:35) hard for Solomon to defend Marcus but (00:39:37) the problem is like Marcus would have (00:39:39) needed 5 10 years to fully mature and (00:39:42) when you're a public company that's hard (00:39:43) to defend I don't know maybe Solomon (00:39:45) doesn't agree and he's like I did the (00:39:47) right thing and I change it but like in (00:39:48) my opinion he should have stick the guns (00:39:50) to doing that. Jamie is now doing neo (00:39:53) bankanking is going in but he's going in (00:39:55) a bit later now. Um and so you will see (00:39:57) different banks I think some of them (00:40:00) will uh reinvent themselves become neo (00:40:03) banks become techled use AI to like you (00:40:06) know reinvent themselves some of them (00:40:07) will not and they will with their way (00:40:09) right like which is always what's (00:40:10) happening the disruption of an industry (00:40:12) so it will depend on the leadership of (00:40:14) those financial institutions (00:40:16) >> you said about 2021 and like high (00:40:18) valuations high prices I I remember the (00:40:21) round where you were done at 45 (00:40:23) >> by who it was off bank no (00:40:25) >> uh it's Not entirely true. That's also a (00:40:27) little bit media, but like yes, there (00:40:28) was there was some shares bought at 45 (00:40:30) as well. Yes, that's right. (00:40:31) >> Okay. (00:40:33) Knowing what you know now, are you happy (00:40:35) you did that? And is there anything you (00:40:36) would have done differently? (00:40:40) >> I It's a good question. I think that (00:40:44) when you're in that kind of high growth (00:40:46) phase, one thing to keep a close eye on (00:40:49) is your multiple expansion going faster (00:40:52) than what your revenue is growing. (00:40:54) Right? If your revenue is growing faster (00:40:56) than your multiples, then you're (00:40:58) probably fine. But if you start seeing (00:41:00) the opposite where multiples are (00:41:01) expanding faster than revenue growth, (00:41:03) that may potentially be a problem longer (00:41:06) term, right? So I think today maybe that (00:41:09) like I I I could have been, you know, (00:41:12) more careful specifically on the hiring (00:41:14) side because it was very sad and (00:41:16) difficult to me to like a few quarters (00:41:19) earlier be hiring at a high pace and (00:41:21) then a few quarters later have to (00:41:22) announce layoffs. that that I felt that (00:41:24) like I should have predicted and been (00:41:27) more cautious about. Was that the (00:41:29) hardest board meeting? You mentioned the (00:41:30) board meeting earlier where I was like, (00:41:31) "Oh, that was kind of kind of nice and (00:41:32) easy with the expansion." You know, (00:41:34) you've got an amazing board. (00:41:36) >> Uh and we're not talking about the board (00:41:37) in any other capacity. I'm just talking (00:41:39) about like a hard board meeting. Does (00:41:40) Mike ever get angry, by the way? (00:41:42) >> No, (00:41:43) >> he doesn't. (00:41:43) >> No. Angry is not the word for my (00:41:45) >> I would be [ __ ] scared if he was (00:41:47) >> No. I mean, it's actually it's it's less (00:41:49) scary because he doesn't get angry. Uh (00:41:51) what what he's like disappointed. (00:41:53) >> Disengaged is the worst. Like the worst (00:41:55) thing you don't want to get with Michael (00:41:56) is disengaged. That's like the big (00:41:58) warning sign. (00:41:59) >> He's never lay on the floor and (00:42:00) pretended to sleep. (00:42:01) >> No, he No, but but you No, I mean, yeah, (00:42:05) he I mean, he's so amazing. I love him (00:42:07) so much. I think he's fantastic, but but (00:42:09) you obviously care a lot to make sure (00:42:10) that like you want to see that he's (00:42:12) engaged. (00:42:12) >> Can I ask you what's been your biggest (00:42:13) lesson from working with him? Cuz you're (00:42:15) like his chosen child in the nicest way, (00:42:17) which is amazing, right? Well done. (00:42:20) Look, I've worked with him so many. It's (00:42:21) a funny story how we ended up working (00:42:23) with him because at that point of time, (00:42:25) uh, Seoa, there was a guy called Chris (00:42:27) Olson, which is amazing, still a friend (00:42:28) of mine who was at Sequoa, and he did (00:42:29) kind of found CLA in (00:42:31) >> drive. No. (00:42:32) >> Yeah, exactly. (00:42:33) >> And he did, uh, the Clana investment. (00:42:35) So, you know, there was like this story (00:42:37) where where like I'm talking to this guy (00:42:40) saying, "Hey, do you because there was a (00:42:41) guy in Stockholm that knew Sequoia a (00:42:42) little bit." So, I was like talking to (00:42:43) him, "Hey, do you would you do you think (00:42:45) that Sequoa could be invested in, you (00:42:47) know, interested in investing in Clana?" (00:42:48) and and he's like, "No, that would never (00:42:50) happen." And I was like, "Okay, fine." (00:42:51) So, we just talked to the European (00:42:52) funds. We're just like, "It's not even (00:42:54) worth, right?" But I was still looking (00:42:55) there in Google Maps. I was looking like (00:42:57) Sand Hill Road. And I was like dreaming, (00:42:59) you know, could I have the chance to (00:43:00) work with these guys? And then suddenly (00:43:02) my phone is buzzing and I'm like, "Oh, I (00:43:04) had a message." I was like, "This is (00:43:06) Chris Olson from Sequoia." You know, I (00:43:08) was like, "Oh my god." I was like, "Uh (00:43:11) oh." So, what I do is like I know like, (00:43:13) "Okay, this is like dating, right? I (00:43:14) can't call back in three days. I have to (00:43:16) wait. I have to wait three days before I (00:43:18) call back. So, I'm just sitting there (00:43:19) like counting the hours like I cannot I (00:43:22) don't want to look too interested. Yeah, (00:43:24) exactly. So, and then I call Chris and (00:43:26) and we take a meeting and they get super (00:43:29) excited and we instantly wanted to work (00:43:31) with Squire. We were like they're the (00:43:32) best. That's how we saw it. Like we saw (00:43:34) this was going to be a huge, you know, (00:43:35) brand uplift for the company. At this (00:43:37) point of time, you know, Daniel Leis, (00:43:38) Spotify is getting all the tech (00:43:39) credibility. Like all the engineers in (00:43:41) stock wants to work with Spotify. Clown (00:43:42) is some boring, you know, invoicing (00:43:44) company. People are like, "What the hell (00:43:46) is that?" Like so we're like we need (00:43:47) tech we need tech credibility right (00:43:50) Sequoa is going to give us that. So (00:43:51) anyways what happens is little bit later (00:43:54) on Sequoa decides to make the (00:43:56) investment. Chris flies to Stockholm and (00:43:57) we had Matt Michael in a in a hotel (00:43:59) breakfast here in London and then and (00:44:01) then we come and and Chris comes and he (00:44:03) makes this beautiful presentation, (00:44:04) right? And the presentation is like (00:44:05) these are the Apple guys in the garage, (00:44:07) these are the Google guys in the garage (00:44:09) and these are the CLA guys in the garage (00:44:11) and you're going to be the next Google, (00:44:12) you know, and we're just sitting there (00:44:13) just like oh my god, we're going to be (00:44:15) Google, you know, like we're just like (00:44:17) we're buying it all. And then I feel (00:44:19) after a while it's like this is not (00:44:21) good. Like we have to do something cocky (00:44:23) here. We can't just like, you know, just (00:44:25) swallow this and be like, "Okay, please (00:44:26) can we work together?" So, when Chris (00:44:28) exits uh the room and he's just about to (00:44:31) leave, I say, "Hey, Chris, just one (00:44:33) thing. If we are genuinely the next (00:44:35) Google, how come you're the only one (00:44:37) from the Sequoia Partnership that's here (00:44:39) in Stockholm today?" Uh, and then Chris (00:44:42) looks at me, "Oh, I'm so sorry. The (00:44:43) other guys couldn't make it." You know, (00:44:45) whatever. You have to remember they (00:44:46) invested, they took a 25% stake at $100 (00:44:48) million valuation, right? (00:44:50) >> They took a 25% stake. (00:44:51) >> Yeah. At $100 million valuation. That (00:44:53) was the deal. They eventually did that. (00:44:54) So, so, so Chris is like, "I'm so sorry. (00:44:56) Um, we could I could make it." (00:44:58) >> And, and he goes into the elevator and (00:45:01) literally 20 seconds later, this is so (00:45:04) impressive. 20 second later, my phone (00:45:06) starts buzzing and it's Michael Morz. (00:45:08) And Michael Morris is like, "Hey, I'm so (00:45:10) sorry I couldn't make the meeting. And (00:45:12) if we get to do this investment, I'll (00:45:15) join the board." So, thanks to me saying (00:45:16) that and being a little bit cocky and (00:45:18) not being so freaking Swedish, um, you (00:45:20) know, we actually got Mike on the board. (00:45:22) And then since then, I've worked with (00:45:23) Mike and gotten to know him. The thing (00:45:25) people don't understand with Mike is (00:45:26) that like Mike has this, you know, I (00:45:29) always think that if for you to be (00:45:30) really good, for you to really (00:45:32) understand the topic, you need to be in (00:45:34) every freaking detail. You need to read (00:45:35) up a lot and and he does. But the point (00:45:38) is it's almost like sometimes I feel (00:45:40) like he can just take this huge mass of (00:45:42) information and he can like without you (00:45:45) know a second of thought he's just that (00:45:48) is important and he just gets that at a (00:45:51) at a level I I other people I don't (00:45:53) interact with like I interact with like (00:45:54) it's just he's just brilliant at that. (00:45:56) It's just amazing. He can like he he (00:45:58) called me in 2019 (00:46:00) in summer. I don't know why he called me (00:46:02) but he was like and this was when we (00:46:04) were we had been in the US for a few (00:46:05) years. We weren't getting any traction. (00:46:07) the business wasn't doing. And then (00:46:08) Nick, another Nick from Afterpay in (00:46:10) Australia was starting to get, you know, (00:46:12) traction in in the US with buy now pay (00:46:13) later. (00:46:14) >> And Chris just and Michael just calls me (00:46:16) out of the blue and just like, (00:46:17) Sebastian, I think it's now or never. If (00:46:21) we don't do US now, we're never going to (00:46:23) do it. And I was just like, I don't know (00:46:25) how the hell he knew that, but he was (00:46:27) like so spot on. It was exactly the (00:46:29) thing. And I just dropped everything and (00:46:30) I was just like, we have to win the US. (00:46:32) And then I just spent, you know, the (00:46:34) next two years focusing 100% on that. (00:46:37) >> Do Sequoia move the needle for a (00:46:38) company? (00:46:39) >> In my opinion, they do. Yeah, I think (00:46:42) so. I think they do. I think they are. (00:46:44) They I mean, I think that I've been very (00:46:45) impressed with all the people I work (00:46:47) with there. I think they're amazing. And (00:46:48) there's a new generation now with Sonia (00:46:50) and Andrew and and and the new guys (00:46:52) coming. So, like it's really cool. Pat (00:46:54) and (00:46:56) um um yeah, what's his name from Zapos? (00:46:58) Sorry. Alfred. Yeah. um you know so yeah (00:47:01) I think they're amazing (00:47:03) >> when you talk about the expansion of (00:47:05) products and you said obviously about (00:47:06) kind of Afterpay and BNPL god I'm going (00:47:09) to get in trouble for this um I'm (00:47:11) pleased to hear about the movement away (00:47:12) from just pure BNPL because is that not (00:47:15) just evidence that for all the people (00:47:16) that said landing is a shitty business (00:47:18) to be in they were right (00:47:20) >> landing what does that mean (00:47:22) >> well like BNPL is a shitty business why (00:47:24) would it be a shitty business (00:47:25) >> oh it's really hard to build like a a 20 (00:47:28) 30 billion business on BMPL like (00:47:30) consumer landing is a hard business to (00:47:32) make a lot of money. (00:47:34) >> It's like a hard business. Yeah. When I (00:47:35) get a startup pitch me consumer landing, (00:47:37) I'm like, (00:47:39) >> "Sorry, (00:47:41) >> look, you're like, "Wow, he's honest." (00:47:43) >> I think the way I thought about this is (00:47:45) like when we started Clana 20 years (00:47:47) back, we were just like, "Okay, look, (00:47:50) why are we not just making these banking (00:47:51) products better working online?" Because (00:47:53) you got to remember in '05, I mean, the (00:47:56) bank's internet offering were [ __ ] (00:47:58) >> dude. 26. That's all [ __ ] (00:48:01) >> So, it was just like they're all [ __ ] (00:48:02) So, we were just like, "Okay, we're (00:48:03) going to take some of the stuff that the (00:48:04) banks do and we're just going to do it (00:48:06) better online." We did that and then (00:48:08) that grew and we were successful. I (00:48:09) mean, people don't know this, but we (00:48:10) were we we we raised $60,000 in our (00:48:13) first angel investment. $30,000 of that (00:48:15) was spent and then we became profitable (00:48:18) and we were running this as a profitable (00:48:19) company from ' 05 to 19. We had almost (00:48:23) 10 consecutive years of high growth and (00:48:26) profitability that actually got us like (00:48:28) this reward because we were like I think (00:48:31) o only one of two companies in Sweden (00:48:32) that ever had such a long streak of high (00:48:35) growth and profitability at the same (00:48:36) point of time. So the thing is that uh (00:48:39) but what we realized also over time was (00:48:42) suddenly I'm sitting one evening and I'm (00:48:44) looking at my P&L and I'm like wow what (00:48:46) is that line? Oh [ __ ] that's late fees (00:48:49) right? that's a big revenue line. And I (00:48:52) was like, that's not going to be (00:48:53) long-term sustainable, right? So, at (00:48:55) some point of time, I started thinking, (00:48:56) wow, you know what? We're actually doing (00:48:57) lending. What does that mean? What does (00:48:59) it mean for consumers? What does it mean (00:49:00) for their financial life? You know, (00:49:02) what's the implications of this? Um, and (00:49:05) at that point of time, I I thought to (00:49:07) myself, there's two things I can do (00:49:09) here, right? I can either sell this (00:49:10) business and say, oh [ __ ] we're making (00:49:12) a little bit too much on interest and (00:49:14) and and late fees. Let's go and do (00:49:16) something else, you know? or I can try (00:49:18) to change this. And uh one of my (00:49:20) co-founders left at that point of time, (00:49:22) right? But I said, "No, I'm going to (00:49:24) stay and I'm going to make the change." (00:49:25) And I realized that the buy now pay (00:49:27) later credit offering is a healthier one (00:49:30) than your credit card. Like you on your (00:49:32) credit card, you put all your spending (00:49:34) full month on that and then the bank (00:49:37) tries to push you to revolve. You build (00:49:38) up a balance of a few,000 or dollars and (00:49:41) then you pay very high interest. So I (00:49:43) was like, but I remember when I worked (00:49:44) at Bur King, it used to be like press (00:49:46) one for debit, press two for credit when (00:49:48) I would swipe my card. And so like where (00:49:51) did that go? Well, banks didn't like it (00:49:53) because the problem was if you were (00:49:54) pressing debit now and then your bill at (00:49:56) the end of the month was much smaller (00:49:58) and you were less likely to revolve and (00:50:00) borrow money and they would make less (00:50:01) money. So they removed the debit button, (00:50:03) right? And I was like, "No, no, no. (00:50:04) Let's bring that back. Let's make sure (00:50:06) clown offers press one for debit. 20% of (00:50:08) our transactions are debit and then the (00:50:10) rest is credit but the credit is (00:50:12) interest free fixed installments no (00:50:14) revolving revolving we've removed it (00:50:16) cost us we gave up $100 million of (00:50:18) revenue when we removed revolving (00:50:20) because we used to do it in Nordics we (00:50:22) took it away uh we even didn't have late (00:50:24) fees in the UK for a period of time at (00:50:26) all but it turned out that that wasn't (00:50:28) great either because then people to some (00:50:29) degree were overextended themselves so (00:50:32) it's good to have a little bit of fee (00:50:33) some kind of consequence of not paying (00:50:34) on time so we we got that back but you (00:50:37) have to be mindful of not making too (00:50:38) much money on it because people will (00:50:41) tend to use it in a way that's not good (00:50:43) for them. So you just have to find the (00:50:45) balance. But over the years we iterated (00:50:47) on a model that we feel like is a better (00:50:49) alternative to credit cards. If 10 years (00:50:51) from now less people have credit cards (00:50:52) and more people have debit cards, we and (00:50:53) then use buy now pay later occasionally, (00:50:55) I would argue it's a better society. (00:50:56) Like that's my belief, right? And that's (00:50:58) what we've been pushing. And the (00:51:00) consumers now we see they are they love (00:51:02) that. They reward us for that. They they (00:51:04) they agree with that, right? And that (00:51:06) doesn't mean you're still in credit. So (00:51:07) you're still going to have unfortunately (00:51:09) occasionally people who overextend (00:51:10) themselves. You have to be mindful about (00:51:11) that. You have to how you help them when (00:51:13) they're distressed and so forth. It's a (00:51:14) difficult business in that sense. It's a (00:51:16) bank, right? So you have to be mindful (00:51:17) about these things. But generally (00:51:19) speaking, the type of product we're (00:51:21) offering is a better product than the (00:51:22) traditional products of the banks. (00:51:23) >> You mentioned starting with Spotify and (00:51:26) Cler in Stockholm together. And (00:51:28) obviously Stockholm's been the (00:51:30) birthplace of great AI companies in the (00:51:32) last year with Lagora and Lovable to (00:51:34) name a few. Um, I'm a young 18-year-old (00:51:37) Stockholm entrepreneur. Seb, you've got (00:51:39) this incredible experience and you've (00:51:41) been to the US and oh, how wonderful. Do (00:51:43) I have to be in the US if I want to (00:51:45) build a big startup today? (00:51:47) >> No. (00:51:49) >> H. (00:51:50) >> Do you disagree? (00:51:51) >> No. [ __ ] I wouldn't be doing project. (00:51:54) But every US experienced founder tells (00:51:59) them yes, you do. (00:52:00) >> Yeah. I think I mean to some degree from (00:52:02) a European perspective it's obviously (00:52:03) partially sad to see that a lot of the (00:52:05) successful AI and companies in the US (00:52:08) where actually their the founders are (00:52:11) European (00:52:12) >> 100% (00:52:12) >> right it's a bit sad but that's just how (00:52:14) it is I think that the I mean I remember (00:52:17) true caller are friends of mine also (00:52:19) fantastic company European (00:52:21) >> I was at when they did that (00:52:23) >> yeah Nami and Zai and they were like (00:52:25) told by the VCs the American VCs like (00:52:27) you have to move your engineering center (00:52:30) to Silicon Valley because otherwise (00:52:32) everything's going to go to [ __ ] And (00:52:34) they did and it was a disaster and they (00:52:37) had it for a year. You know, they tried (00:52:39) to recruit but obviously nobody in (00:52:40) Silicon Valley knew what like true color (00:52:42) was and they were already a pretty grown (00:52:44) company so they couldn't like attract as (00:52:45) a startup in that sense and they had a (00:52:47) hard time recruiting and whatever and (00:52:49) they just like why are we doing this and (00:52:50) then they shut it down. They lost a year (00:52:51) on that. They lost a year on that. Lost (00:52:53) so much traction and then some of these (00:52:55) VCs were like oh you're not doing well. (00:52:58) So now we're not going to put our (00:52:59) partner on your board anymore. We put (00:53:00) some junior guy on your board because we (00:53:01) don't care because you're not our top (00:53:02) priority of companies anymore. (00:53:03) >> It's the most savage indictment, isn't (00:53:05) it? When you get the associate join the (00:53:06) board and you're like, oh [ __ ] I've (00:53:07) been relegated. (00:53:08) >> Exactly. So first they're advising them (00:53:10) to do this thing and then when they (00:53:12) actually go and execute it and it turns (00:53:13) out to be a disaster, they're like, I'm (00:53:14) sorry, your company is not that good. (00:53:16) Like I mean it's just terrible, right? (00:53:18) It was terrible. And I think that like (00:53:20) >> Do you think the state of VC is very (00:53:21) good today? Actually, (00:53:23) >> I think again it's changing so fast. I (00:53:25) think that the the the challenge right (00:53:28) now is a lot of them are piling money (00:53:29) into AI that they don't necessarily (00:53:31) fully understand like how good is it? (00:53:34) How differentiating is it? Is it really (00:53:36) does it have a real note? (00:53:37) >> Well, what should I know then as one of (00:53:39) these VCs piling money into AI? (00:53:42) >> Well, I think what you should be doing (00:53:43) is coding with cursor and building (00:53:45) things yourself. If you do that, I think (00:53:47) you do that if I remember correctly. (00:53:48) >> Yeah. But lovable. (00:53:49) >> Yeah. Yeah. Yeah. Well, you only lovable (00:53:50) or only have you tried cursor as well? (00:53:52) >> I've tried core code. (00:53:53) >> Okay, good. Yeah. So the point is like I (00:53:55) just think that like I if I meet (00:53:57) investors today that haven't actually (00:53:59) downloaded and tried to build something (00:54:01) themselves, I think they don't have the (00:54:03) skill set to make an evaluation of the (00:54:05) company they're looking at. I think it's (00:54:06) so critical to actually just understand (00:54:09) how powerful these tools are today (00:54:10) before you make those decisions. If you (00:54:12) then if you have that insight, if you (00:54:14) understand that and then you make your (00:54:16) decisions, fine. Like there's going to (00:54:17) be opportunities. (00:54:18) >> I think Castle will lose half of its (00:54:19) revenue in 2026. (00:54:21) >> Is that your prediction? Yeah. Why is (00:54:22) that? Because claw code's just eaten (00:54:24) their lunch. I don't see any engineering (00:54:26) team that's still on cursor and that's (00:54:27) >> we love it actually. We use it all the (00:54:29) time. (00:54:29) >> Really? (00:54:30) >> Yeah. Is that because you have an (00:54:31) enterprise deployment and contract? (00:54:33) >> No, I don't think so. I don't know why. (00:54:34) Like so I kind of switch between claw. (00:54:36) I'm a big entropic fan. I love I love (00:54:38) claw the chat version of it as well. I (00:54:40) use that all the time. I just find that (00:54:41) like it depends on kind of the task. So (00:54:44) sometimes I use I'm almost using them as (00:54:46) like I would go into cursor, I would (00:54:47) write some things, then I go to cloud (00:54:49) code, then I'll ask cloud code to do (00:54:51) some other things. I still feel they (00:54:52) have like almost distinct personalities (00:54:53) and skills and so I kind of enjoy still (00:54:56) and and I need an ID. The problem is (00:54:57) also like because I wasn't an engineer I (00:54:59) never used like you know uh VS code or (00:55:02) any of these tools. I still need an ID (00:55:04) today. So like cursor is my standard ID (00:55:07) right just like as a as a consequence of (00:55:08) that. So you know you could you could be (00:55:11) right but I'm actually more optimistic (00:55:12) about their future than that. (00:55:13) >> You can invest in Anthropic at 360 or (00:55:15) open AI at 500. I'm giving the discount (00:55:18) to make it easier. (00:55:19) >> Don't make me answer this question (00:55:20) please. like (00:55:23) >> I think I think they're going in very (00:55:25) different directions. (00:55:26) >> I think that if you if at least my (00:55:28) experience with Open AI is that it's (00:55:30) becoming a consumer company and if (00:55:33) you're if I'm building AI for like a (00:55:36) billion people, right? I'm building AI (00:55:38) that I would focus on making sure that (00:55:40) like for example, there's going to be (00:55:42) people that are going to seek AI as a (00:55:44) friend, as as a friend in their (00:55:46) day-to-day life, as somebody more like (00:55:48) from that movie. What's the movie again? (00:55:49) Uh uh (00:55:51) >> the Scarlet your handsome one. (00:55:52) >> Yeah. What's it called? (00:55:53) >> Um this (00:55:55) >> her. Exactly. Right. So some people (00:55:57) going to more look for the her (00:55:58) experience and and I think if you know (00:56:01) that's chat that's chatb to me because (00:56:03) if I I'm so big I'm such a big consumer (00:56:05) brand. I'm going to start looking at my (00:56:07) KPIs and I'm going to optimize for (00:56:09) emotional connection for people wanting (00:56:11) to like you know how much time are they (00:56:13) spending with my product? (00:56:14) >> Yeah. Is that really going to be a chat (00:56:16) GBTR or is that going to be a companion (00:56:17) product and there's 15 provider which is (00:56:19) so specialized (00:56:21) >> it could be but my point is that like if (00:56:22) I'm just looking at who their audience (00:56:24) is today it's very likely that they will (00:56:26) optimize for that relationship aspect of (00:56:28) like being your counselor your provider (00:56:30) your your play friend that you're (00:56:32) playing with or playing games with or (00:56:34) whatever right I mean Claude to me is (00:56:36) very different Claude is like my (00:56:39) intelligent advisor and and I I'm trying (00:56:41) to push and I say to Entropic all the (00:56:43) time like I don't want an AI that tell (00:56:45) me you're so great, man. Like, I (00:56:47) understand that that's like a nice (00:56:48) experience and some people may enjoy (00:56:50) that. Like, I I actually enjoy somebody (00:56:51) just telling me how awesome I am every (00:56:53) day. Personally, I'm not that interested (00:56:54) in that part. What I'm interested in is (00:56:56) rather somebody's telling me, "Sbastian, (00:56:58) that's freaking stupid." Like, don't do (00:57:00) that. That makes zero sense. Like, I (00:57:02) want an AI to tell me you're wrong, man. (00:57:05) Like, don't do that. That makes no (00:57:07) sense. And I feel currently Claude is (00:57:09) more it's more likely that Claude (00:57:12) provides me like less biased. It's it's (00:57:14) trying less to please me. And I think if (00:57:16) you're building a product open AI the (00:57:18) risk is you becoming into the pleasing (00:57:19) because I think a lot of people like (00:57:21) that they and I I would I mean I (00:57:22) understand if I would if I would use AI (00:57:24) for entertainment then also I want to be (00:57:26) like I want to be entertained. I want to (00:57:28) feel pleased. I want to feel happy. I (00:57:29) want to feel you know it's a different (00:57:30) product. See what I mean? It's just a (00:57:32) different thing you're looking for. Uh (00:57:34) and and I have less interest in that. I (00:57:36) have more interest in somebody telling (00:57:37) me you're totally off. Don't do that. (00:57:40) >> So would rather be anthropic. (00:57:41) >> Yes. For that perspective. Yes. (00:57:43) >> Yeah. 150 or $149 billion. (00:57:46) >> I didn't look at the valuation. I didn't (00:57:47) billion was the revenue expectations by (00:57:49) 2030 (00:57:50) >> Oh, really? (00:57:50) >> Yeah. (00:57:51) >> Which was phenomenal. (00:57:53) >> Yeah. (00:57:53) >> And impressive. Um you invest now a lot (00:57:57) >> as well through flat. No. (00:57:59) >> What have you changed your mind on since (00:58:00) also being an investor? (00:58:02) >> No, I have changed my mind on software. (00:58:04) Right. So I think that uh software is (00:58:08) getting much more risky uh and it's much (00:58:10) more unclear uh what the future of SAS (00:58:13) is. So generally I try to we we for (00:58:16) example make made a big flat made a big (00:58:18) investment in defensor which is like a (00:58:21) uh you know into um military arms you (00:58:24) know type of thing not arms but military (00:58:26) defense. (00:58:27) >> Yeah. (00:58:27) >> And um and that's very much like not you (00:58:30) know there's some drones in there and (00:58:32) there stuff like that but it's not it's (00:58:33) very different. It's not SAS, right? (00:58:35) >> I think data center is the one as the (00:58:37) most underinvested categories today. (00:58:39) When you look at inference needing to (00:58:41) run for 24 hours a day for most of the (00:58:42) knowledge worker population and it (00:58:44) running for like 1% of knowledge worker (00:58:46) population today, I'm like, (00:58:48) >> why the [ __ ] is more money not going (00:58:50) into data centers? (00:58:51) >> Do you agree with that? And do you think (00:58:53) >> I I I have a very It's funny you asked (00:58:55) me about this. I've actually play I play (00:58:57) around sometimes when I have time. I (00:58:58) play around with Suno. Have you played (00:59:00) around with Suno? (00:59:00) >> Dude, I love Suno. I actually I actually (00:59:02) sent Nick at Revolute a song to get him (00:59:06) to come on the show most recently and it (00:59:07) helped get him on the show. Yeah, that's (00:59:09) awesome. (00:59:09) >> It helped get him on. Yeah, (00:59:11) >> I love So. So, what I've done with Soon (00:59:12) is like I was I was playing around and (00:59:14) I've been doing some songs. I even (00:59:16) published it on Spotify for the fun of (00:59:17) it. So, you can go to ClaB. I have like (00:59:19) 74 monthly listeners or something. (00:59:23) Uh and my kids get so annoyed when I (00:59:25) publish songs in my name. Um it's really (00:59:27) funny. But uh two of those songs are (00:59:30) called compression. (00:59:31) And this came from a conversation I had (00:59:33) with Claude. So I blame all the lyrics (00:59:35) on Claude. Don't blame him on me. I was (00:59:37) only the producer. Claude wrote the (00:59:38) lyrics. Then I had to give Claude some (00:59:40) artistic, you know, artistic freedom. Um (00:59:43) I'm very interested in this. Like I got (00:59:45) this question. I was on on a a (00:59:46) conference in Yellowstone and I was on (00:59:50) the stage crazy enough with Sam Alman (00:59:53) and Eric Schmidt. And from the audience (00:59:55) comes this question. How is it possible (00:59:58) that you can take the whole of Chach5 as (01:00:01) an example, one of these models once (01:00:03) they've been trained, once the training (01:00:04) is over and the whole thing is like (01:00:06) done, and fit it on a USB stick? (01:00:10) How's that possible (01:00:12) that it's not bigger in size, it's just (01:00:13) like a a few hundred gigabytes or (01:00:15) whatever the size of the models are? (01:00:17) And uh and then they gave like different (01:00:20) answers and I had an answer in my head (01:00:21) but I felt embarrassed in that in that (01:00:23) in that setting to say and I think what (01:00:25) people underestimate with AI it's a (01:00:27) compression technology. So what that (01:00:30) means is if you historically put data in (01:00:32) a database right you say a database (01:00:34) record okay uh Clana has a customer (01:00:36) called Sephora and then we write again (01:00:38) CLA has a customer in Sephora you (01:00:40) created this tremendous amount of (01:00:41) duplication. If you look at any large (01:00:44) enterprise company, they will have the (01:00:45) same information over and over and over (01:00:49) again, right? But if you look at (01:00:51) Wikipedia, how many articles is there (01:00:53) about Clana? One. Why aren't there 15? (01:00:56) What do they do so magically? How can it (01:00:58) be that Cla historically had a customer (01:01:00) relationship with Sephora and we had (01:01:02) information about that customer (01:01:03) relationship in Slack, in Salesforce, in (01:01:06) Google Docs, in Google Slide? Kind of (01:01:08) the same information over and over (01:01:10) again. But on Wikipedia, it's just one (01:01:12) article. How do they do that? And I (01:01:14) realize when you train the model, you (01:01:17) give it if I tell that like Harry uh you (01:01:21) know is not only running a fantastic (01:01:23) podcast but also runs a VC that like if (01:01:25) you tell it tell it once when you train (01:01:27) it, it will forget it. It will ignore (01:01:29) that information. But if you tell it (01:01:30) enough number of times, it will remember (01:01:32) it and then when you go and ask it, it (01:01:34) will know that information. But it's not (01:01:36) storing it twice because if it's getting (01:01:38) the same information that it already (01:01:39) knows, it doesn't move the tokens. So, (01:01:41) it's automatically compressing all the (01:01:44) information. This is why you can take (01:01:46) the whole freaking internet, all human (01:01:48) knowledge, and compress it down to a few (01:01:51) hundred gigabytes. That's a huge amount (01:01:54) of information. Now, obviously, you lose (01:01:56) precision. This is why it's very (01:01:57) worthless to go and ask, "What is the (01:01:59) opening hour of the Starbucks down on (01:02:00) the corner?" Like, the AI would be like, (01:02:02) "I have no clue." the equivalent of (01:02:06) chatb5. If I asked this AI, so AI is (01:02:08) responsible if I'm wrong, but the chatb5 (01:02:12) as an example, the equivalent of the (01:02:14) number of gigabytes that that model is (01:02:15) is the equivalent of about 2 three days (01:02:18) of weather data from the whole globe. (01:02:20) That's it. So how so how can it then be (01:02:22) so capable of answering all these (01:02:24) questions? Because unfortunately, (01:02:27) despite what we humans would like to (01:02:29) say, the amount of true novel (01:02:32) information and knowledge in human (01:02:34) society is quite limited. What we see is (01:02:37) repetitions and variations on the same (01:02:40) themes over and over and over again. So (01:02:43) the point being to answer your question (01:02:45) when we realize this we also realize (01:02:48) that we're compressing knowledge at an (01:02:50) ex in that's why people are playing (01:02:52) around doing this like oh look at my (01:02:53) iMac mini I was running my own model on (01:02:55) it it actually works like I can run this (01:02:56) on a raspberry you know like the thing (01:02:58) is un like Romeo and Juliet that story (01:03:02) exist in a hundred different variations (01:03:04) and the unfortunately thing is when you (01:03:05) compress it down with math it knows AI (01:03:08) sees that not as Romeo Juliet and then (01:03:11) another love story another love story it (01:03:13) is a love story and then knows it has (01:03:15) slightly different names in different (01:03:17) variations. So, it's a massive (01:03:20) compression. And now comes the question, (01:03:22) do we need all of that compute in the (01:03:24) future? And I am, you know, I happen to (01:03:26) have this amazing conversation with (01:03:28) Michael J. Bur about this the other (01:03:29) week. (01:03:29) >> Wow. (01:03:30) >> Because I was talking about this because (01:03:31) he's doing those bets that it's not (01:03:33) right. And and I said, look, I I I think (01:03:37) there's two arguments for and against. (01:03:40) One is if you look at enterprise, (01:03:44) enterprise, what does enterprise want? (01:03:45) Enterprise wants highest quality at (01:03:47) lowest cost. If why would I recomputee (01:03:51) my information about Sephora over and (01:03:53) over again? If somebody could help me to (01:03:55) compress that down to just one single (01:03:56) source of truth and not having all that (01:03:58) information unnecessary, I'll take it. (01:04:01) I'll save a lot of money. I don't need (01:04:02) that all that compute. Why do I want to (01:04:03) do that? So, in enterprise, you're going (01:04:05) to see a dramatic squeeze and (01:04:06) compression. The counterargument to this (01:04:09) is you and I then go out and say, "Hey, (01:04:12) we've had this amazing podcast. Now, (01:04:13) let's watch a movie together. We want to (01:04:15) watch Star Wars but with our faces. So (01:04:17) you'll be Darth Vader and I'll be Luke. (01:04:18) Right now that needs generation. That (01:04:21) needs a you know a data center to (01:04:23) generate that for us. So the question is (01:04:25) just like what power will be greater the (01:04:28) compression of enterprise data or the a (01:04:31) generation of new stuff for (01:04:32) entertainment and other things. And I (01:04:34) don't know the answer to that question. (01:04:36) I think you can argue both ways. But in (01:04:37) enterprise data there's going to be a (01:04:39) massive compression that's going to come (01:04:41) naturally through this technology. And (01:04:43) that's why I'm a little bit like I'm and (01:04:45) I don't want to be the guy who said (01:04:46) there's only going to be four computers (01:04:47) in the world, you know, like they bit (01:04:49) they like I don't want to be that guy. (01:04:50) So I want to be a bit mindfully (01:04:51) >> clips stay forever. Don't do this to (01:04:53) yourself. (01:04:54) >> He said there was only going to be (01:04:55) >> in 2040 they're going to play this clip. (01:04:57) I was like, can you believe it? (01:04:58) >> He was so stupid. No, but like but I'm (01:05:00) just saying that like it will so I I (01:05:02) don't know which power is going to be (01:05:03) greater, right? Like which one? (01:05:04) >> If you run with that enterprise (01:05:06) compression, what does that mean for (01:05:08) data centers and for chip companies like (01:05:10) Nvidia? Well, that would mean that you (01:05:12) would mean signific you would need (01:05:15) significantly less. That's the (01:05:17) consequence because you because the AI (01:05:19) the problem is why why do we have so (01:05:21) much big software? Why do we have so (01:05:23) much data? It's all a mess because we (01:05:26) had humans who were trying to do their (01:05:28) best but the right hand didn't know what (01:05:30) the left hand were doing and they were (01:05:32) overwriting that and the code over there (01:05:34) didn't there and then people start doing (01:05:35) transformations and stuff and it all (01:05:37) becomes a big mess and actually you know (01:05:39) Wikipedia is the most successful (01:05:41) knowledge graph in the world I would (01:05:43) argue but if you look at the standards (01:05:45) that they apply how do they do that (01:05:47) they're very like one very I'll give you (01:05:48) a very simple like I love this principle (01:05:51) so if you go to Wikipedia and you try to (01:05:53) create a new article about a new topic. (01:05:55) It's very hard. Like if you go to Google (01:05:57) Docs, you just click new. Boom. You (01:05:58) start writing a new document. You know, (01:06:00) if you go to cursor, you start new, new (01:06:01) codebase. Wikipedia, where is the new (01:06:03) button? There is none. Do you know how (01:06:05) to find a new button? (01:06:07) No. You have to search for an article. (01:06:10) And if you search for something that (01:06:12) doesn't exist, then you're allowed to (01:06:14) create new. See what I mean? And that's (01:06:16) so different. And companies doesn't work (01:06:18) like that. In companies, like, hey, I (01:06:20) have an idea. I think clone should be (01:06:21) doing this. Like, okay, let's just start (01:06:22) coding. But let's go and check. Maybe we (01:06:24) already have code doing this. Maybe we (01:06:27) already do that. So the point is that (01:06:29) like the reason that compression hasn't (01:06:31) happened historically is just because so (01:06:33) many people have been involved with (01:06:34) different experience, knowledge and so (01:06:36) forth. And it just creates this massive (01:06:37) mess and nobody has been as like, you (01:06:40) know, disciplined as Vicky Pedience has (01:06:42) been to like keep it to one source of (01:06:45) truth and so forth. They really been (01:06:46) amazing at that. You can read up a lot (01:06:48) about how they've been doing, but AI is (01:06:49) going to help with that. AI is going to (01:06:51) help organizations say, "Hey, should we (01:06:52) really be doing this thing because we (01:06:54) already have code for this and so (01:06:56) forth." It's not there yet, but that's (01:06:57) going to happen. Not because AI gets (01:07:00) smarter, because it's economically sound (01:07:03) in a business. It's not economically (01:07:05) sound to duplicate. It's economically (01:07:07) sound to reuse what you already have. (01:07:09) That's the reason it's going to happen. (01:07:11) It's not because like, you know, this is (01:07:13) like some hypothesis. It's just like it (01:07:15) makes sense economically to not do what (01:07:17) you've already done again, right? Were (01:07:20) there any other takeaways from your (01:07:21) conversation with Michael Barry? He's a (01:07:24) phenomenal thinker. (01:07:25) >> Yeah. Yeah. No, we enjoy I think I mean (01:07:26) in this in this regard I think we were (01:07:28) lined, right? Like but it was funny (01:07:30) because we were talking about this you (01:07:32) know how much novelty is there really. (01:07:34) Right. And it's an interesting concept (01:07:35) because I as I said to him also there's (01:07:37) other data that suggests otherwise. I've (01:07:39) heard data that suggest that 30% of the (01:07:41) searches on Google every day are new. (01:07:42) Like I don't know if that's true. It (01:07:44) sounds crazy to me but there are some (01:07:46) data to suggest. So maybe there is more (01:07:47) novelty. And I think that's like I don't (01:07:49) know the answer to these questions, but (01:07:50) I think this is like very very (01:07:52) fascinating to me to like see uh the the (01:07:55) counter effects of these. (01:07:56) >> Before we go into a quick fight, it's (01:07:58) been not knowing the answer to the (01:07:59) questions. I'm sure you're in CEO (01:08:01) groups, go to CEO events, sit in green (01:08:04) rooms, and there are topics or themes (01:08:06) that CEOs discuss about AI that they do (01:08:10) not discuss publicly. (01:08:11) >> What do you think they most are? Well, I (01:08:14) think most of them as I said previously (01:08:15) with me and Dario is that I think most (01:08:17) of them recognize in my first of all I (01:08:19) don't go to that many of those to be (01:08:20) honest because I'm hard coding and (01:08:22) working with my teams and trying to make (01:08:24) sure that client is as successful as (01:08:26) possible through this transformation. So (01:08:27) I just I spend very little time on those (01:08:29) kind of you know things. (01:08:30) >> Does this generation make every CEO even (01:08:33) public company CEO is a builder again? (01:08:35) >> I think it has to be. Yeah. (01:08:37) >> Doesn't it? (01:08:37) >> Well, we were again I was with Mike last (01:08:39) night from Atlass and he's I'm up at 5 (01:08:41) a.m. coding. M (01:08:42) >> I'm like was that the same before? He's (01:08:44) like no (01:08:44) >> no. Well, I think so. But it's also (01:08:46) amazing, right? It's acceler it's I mean (01:08:48) I think for somebody like myself who (01:08:49) didn't used to code or couldn't code (01:08:51) like I think it's fantastic to to be (01:08:54) able to take my ideas and thoughts and (01:08:56) turn them into something I can show (01:08:58) others. It doesn't mean to have to be (01:08:59) production ready, but I can articulate (01:09:02) things at a very different level of (01:09:04) quality than just like trying to explain (01:09:06) something on a whiteboard or you know (01:09:08) whatever. Now I can actually bring (01:09:09) things to people. Uh I had this very (01:09:11) very unique experience with Claude just (01:09:13) last week where I was we were trying to (01:09:15) talk about to kind of communicate a very (01:09:18) specific thing that was like touching on (01:09:21) you know accounting and finances and (01:09:24) predictions and stuff like it was pretty (01:09:25) like very very complex thing and we were (01:09:27) just talking about it and this was like (01:09:29) the first time I went to I went to (01:09:31) Claude and I said hey can we like (01:09:32) interact a few times I want to like see (01:09:33) if we can like explain this concept and (01:09:36) after a few iterations I got this like (01:09:39) beautiful animation wasn't even a a (01:09:42) slide because it was an HTML file, but (01:09:44) it was just like and I I thought to (01:09:46) myself, wow, you know what? This is (01:09:48) actually the first time I felt that AI (01:09:50) could do something that humans couldn't. (01:09:53) And and it was the first time I had that (01:09:54) experience. And the reason was because (01:09:56) if I would have liked to do the same (01:09:58) animation and beautiful pedagogical (01:10:00) explanation of a very complex technical (01:10:02) accounting finance thing, historically I (01:10:06) would have bring brought in an animator, (01:10:08) a designer, an accountant, a finance (01:10:12) Google sheet guy, etc. And each one of (01:10:14) them may have been extremely skilled at (01:10:16) what they do, but didn't necessarily (01:10:18) know what the other person needed in (01:10:21) order to to perfect that outcome. And so (01:10:24) the animator would have been like, (01:10:25) "Yeah, I can animate like this and (01:10:26) that." But they don't really understand (01:10:27) what they're animating, right? The (01:10:28) financial concept. The financial guy may (01:10:30) be like, "I can do the numbers like this (01:10:31) and that, but I don't really understand (01:10:32) why we would need this like animation (01:10:34) cuz I read the numbers and I get (01:10:35) everything. I don't need this freaking (01:10:36) visualization, right?" So like, so the (01:10:38) point but here Claude Claude had all the (01:10:41) skills in one and then created this and (01:10:44) I was like, I could not aim would not (01:10:46) have done that. Like it and it was just (01:10:48) like such a experience to There's a (01:10:50) special moment when AI exceeds human (01:10:51) capabilities. (01:10:52) >> And I think in this in this experience (01:10:53) to me it did like it did something that (01:10:55) I I don't think a sing it's not like a (01:10:57) single person could have done each part (01:10:58) of different parts. But what was rec (01:11:00) like how many people are you going to (01:11:01) find in the world that are great (01:11:02) animators, great visualization people, (01:11:05) super pedagogical, knows everything (01:11:07) about financial accounting, understand (01:11:08) financial services. Like you know how (01:11:10) who you going to find (01:11:11) >> rare van diagram. (01:11:12) >> Yeah. Exactly. Like it's going to be (01:11:14) hard to find the person that's good at (01:11:15) all of these things in one. Right. (01:11:16) >> Final one for a quick fire. What happens (01:11:18) to Elon with Grock? I would never dare (01:11:20) to bet against is when you have two (01:11:22) people you would never dare to bet (01:11:23) against in the same market. And most (01:11:25) people would never bet dare to bet (01:11:27) against Sam. (01:11:28) >> Mhm. (01:11:30) >> And so if you'd never dare to bet (01:11:32) against Sam, but you'd also never dare (01:11:34) to bet against Elon, one of your prior (01:11:36) assumptions has to be wrong. (01:11:38) >> Yeah. I I think he's going to do really (01:11:40) well. I I mean I think it's if you think (01:11:42) about the fact that he could in a few (01:11:44) weeks in a few weeks put together a (01:11:48) frontier model at that quality level. I (01:11:51) mean you got to give the credit to the (01:11:52) guy like I mean I that's just like it's (01:11:54) it's insane. Do you use it ever? I (01:11:56) actually, you know what I love about it (01:11:58) is I think he's, and this is funny (01:12:00) because this is exactly what Elon has (01:12:02) been saying all the time and nobody (01:12:03) wants to give him credit for this, but (01:12:04) the point is (01:12:06) Onyx people spread all these rumors like (01:12:09) we were saying like yesterday was a (01:12:10) rumor that like Cla like company offers (01:12:13) buy now pay later for rent. (01:12:14) >> Yeah. Yeah. You're doing rent. (01:12:15) >> Yeah, we're doing rent. And we're like, (01:12:17) "No, we're not doing rent." And nobody (01:12:19) cares, right? But one thing that makes (01:12:21) >> Yeah. It's too late. And people are just (01:12:22) like, "It was Cla like next." And you (01:12:24) know, now they're disappointed. You (01:12:26) don't do rent. (01:12:26) >> Exactly. And then next tweet is Clonado (01:12:29) does rent and then we're trying to like (01:12:30) our comm's team's jumping on and trying (01:12:32) like but but like it's what if it was (01:12:34) established media they would fix it but (01:12:36) like these expost and Tik Tok post like (01:12:38) nobody just it just goes one answer (01:12:40) right but then people write at Grock is (01:12:43) this true (01:12:44) >> and Grock actually almost always answers (01:12:49) correctly and I think that to me is very (01:12:52) very impressive and I think speaks to (01:12:54) Elon's vision of X that over time it can (01:12:56) become the trusted source of information (01:12:59) >> and what he's done with Groipedia where (01:13:00) he's taking inspiration from Wikipedia (01:13:02) what I said and so forth. So I think (01:13:04) that like there's something there and (01:13:06) that's going to be critical for our (01:13:07) societies because right now we have this (01:13:09) scam incams flowing over like AI is (01:13:12) creating you know you know virtual (01:13:14) versions of everything. So I think that (01:13:16) that's to me where the holy you know (01:13:18) where there's something really really (01:13:19) exciting. (01:13:20) >> I agree. Do you know I just wish you (01:13:21) could hide the at rock because I'm too (01:13:24) embarrassed to ask like what is an ideal (01:13:30) like you just explain like quietly I (01:13:33) don't want anyone to see this. Uh dude I (01:13:35) want to do a quick fire with you. So (01:13:36) what have you changed your mind on most (01:13:38) in the last 12 months? (01:13:40) >> I've changed my mind most about the pace (01:13:42) at which the transformation is (01:13:44) happening. I think that I was probably (01:13:46) in the camp of I I thought it was going (01:13:48) to happen too f like faster than it did. (01:13:51) But I think I' I'm always my problem is (01:13:52) that like I overestimate it takes time (01:13:54) for people to change habits and and and (01:13:57) ways of working and stuff like that. So (01:13:58) I actually think it's going to take a (01:14:00) little bit longer the adoption. It's not (01:14:03) necessarily the the capabilities of the (01:14:06) technology, but like how fast people (01:14:07) will adopt it and how it will change. Do (01:14:09) you think that differs for enterprise (01:14:10) versus consumer? Because I I'm actually (01:14:13) the opposite. I'm like surprised by how (01:14:15) quickly consumers adopted it. When you (01:14:17) look at the wow rate, wow numbers for (01:14:20) chat. (01:14:20) >> Yeah, but consumers are always going to (01:14:21) be faster, right? To your point. Way (01:14:23) faster. (01:14:23) >> Yeah, way faster. (01:14:24) >> I think we underestimate how fast (01:14:25) consumers adopt and overestimate how (01:14:27) fast enterprises adopt it. (01:14:28) >> Yeah, for sure. And I think in in kind (01:14:30) of work life, it's definitely it's (01:14:32) definitely going to be slower. (01:14:35) >> What criticism about you stings? Because (01:14:37) it's partly true. (01:14:41) No, I think the criticism that stings is (01:14:43) the one that isn't true, which is well, (01:14:46) I think people sometimes said that like, (01:14:47) oh, he's just trying to make an exit. He (01:14:49) just trying to make a big fast buck or (01:14:51) that like I I just want to, you know, (01:14:54) and after 20 years, I feel like, hello, (01:14:56) you know, haven't I proven now? I'm in (01:14:57) it for the long term. I think the other (01:14:59) one is that like he just he doesn't care (01:15:01) about the consumer. He just wants to (01:15:02) make money on interest rates, etc., (01:15:04) etc., you know, reckless lending and all (01:15:06) that stuff. None of that is true. I I (01:15:08) deeply care about our customers. I (01:15:10) deeply care about them being financially (01:15:12) well off and I think I'm providing a (01:15:14) product that is better than the (01:15:15) alternatives. So like that's that's the (01:15:16) one I stinks because it's just it's not (01:15:18) true. I've never I I don't understand (01:15:20) how you'd ever level that criticism (01:15:22) against you like battling public company (01:15:25) CEO every day. I mean it's like the (01:15:27) opposite of like quick buck there. (01:15:30) Anyway, um what would you do first if (01:15:33) you were not a public company and had no (01:15:35) scrutiny on you? Like if I said here's (01:15:38) an invisibility cloak, you can do (01:15:40) anything, spend any money on anything (01:15:41) internally. What would you do? (01:15:44) >> No, I think the only thing I would do (01:15:46) different is spend less time on (01:15:47) communicating and talking to investors, (01:15:50) right? That's it. Um, I mean to be (01:15:53) honest, but when I look at the strategy (01:15:55) and what I'm trying to do with the (01:15:56) company, I I I I I honestly feel like as (01:16:00) I said that 2015 vision, let's be that (01:16:02) digital financial service assistant. (01:16:04) That's the one I'm executing on. We're (01:16:06) executing on it. The company's executing (01:16:07) on it. I feel we're having tremendous (01:16:10) momentum on it. We see the, you know, (01:16:12) the affection of our customers picking (01:16:13) up our banking products is amazing. (01:16:14) >> Do you think you've told a good story (01:16:16) around that? (01:16:17) >> No, I think communication is hard. I (01:16:19) think (01:16:19) >> because I respectfully I didn't I didn't (01:16:21) know this. I feel terrible. (01:16:23) >> But you and I talked about it even (01:16:24) before this podcast started that I I (01:16:26) think my problem is right is that like (01:16:28) and I and this is definitely what (01:16:30) Michael gives me as feedback like I need (01:16:31) to stick more on message. I need to I (01:16:33) failed probably at this podcast again (01:16:35) and I tried to talk about all kind of (01:16:36) things. I'm just like I love my work. I (01:16:39) love I love what we're doing. I think so (01:16:42) many things are so interesting and (01:16:44) complex and then I sometimes want to (01:16:46) tell people everything. I want to (01:16:47) explain everything to them and I I (01:16:49) should be better at like sticking to, (01:16:51) you know, a clear message. (01:16:53) >> No, totally bad advice. (01:16:56) People buy you, not what you sell. (01:16:59) >> Yeah. (01:17:00) >> Okay. We don't release 30% of shows (01:17:02) because CEOs come on and they just sell. (01:17:05) Let me tell you why our banking products (01:17:07) about it. It sucks. (01:17:08) >> People buy you, not what you sell. (01:17:10) >> I hope so. I hope that over over the (01:17:12) years that will pay off. But I also see (01:17:13) like what's funny is like when we were (01:17:15) the most, you know, uh the most highest (01:17:17) valued fintech in Europe, blah blah (01:17:18) blah, then like I could do no wrong. (01:17:20) Whatever I said was like that's (01:17:22) brilliant, you know, like he's so smart. (01:17:25) And then when we like, you know, when we (01:17:27) came down to 6.5 and we had to do (01:17:29) layoffs, I was like everything he does (01:17:31) is a disaster. (01:17:31) >> Was that a brutal time for you? (01:17:33) >> Of course. Of course it was. It was (01:17:36) >> How do you deal with that? I think you I (01:17:39) mean to me it's also like one story I (01:17:42) was it was I had this very crazy thing (01:17:45) which is I've been doing this for many (01:17:47) years now so I'm a little bit more like (01:17:49) you know thick skin but but what (01:17:51) happened is I'm on this interview with (01:17:52) MSNBC and I've been with an interview (01:17:55) with them many times and generally (01:17:56) speaking like you know questions have (01:17:58) been I would say balanced and fair but (01:18:01) suddenly on this interview they bring in (01:18:03) this new guy on board and this was (01:18:05) exactly when this was happening and I'm (01:18:06) sitting in my home cuz I was doing this (01:18:07) interview for my house and the guy comes (01:18:10) in and he's like this is a disaster. The (01:18:14) company is going to [ __ ] I basically (01:18:15) didn't say that word but like basically (01:18:17) all this you know this is the end of (01:18:18) Clana it's all going down and it's like (01:18:20) and he's just like putting this massive (01:18:22) pressure on me and I'm just sitting (01:18:23) there and I'm I'm almost starting to (01:18:25) crack up and laugh because it was just (01:18:27) like so insane. It's like come on the (01:18:29) business really well like obviously we (01:18:30) need to do some changes but like it's (01:18:32) going to be fine and and I'm trying to (01:18:34) kind of deal with it as professionally (01:18:35) as I can. like no I don't think like (01:18:37) let's blah blah blah as you try to do in (01:18:39) those situations. But the point is I go (01:18:41) out of that I was like oh my god that (01:18:43) was in intense. So I get in my car I (01:18:46) drive to to the office and I was like (01:18:48) there's only one song that I can listen (01:18:50) to now and obviously I put on queen (01:18:53) under pressure. I put it on max volume (01:18:55) and I'm just sitting there under (01:18:57) pressure (01:18:58) and I crack up and the thing is I said (01:19:01) to myself look you know a biggest like (01:19:04) one of the biggest idols of mine (01:19:05) obviously Slat Ibrahimovic born in (01:19:07) Sweden the same day I am 3rd of October (01:19:09) 81 and like and I think of him and I say (01:19:13) I wanted to play Champions League. (01:19:16) How the hell does it feel to go into the (01:19:18) finals and play Champions League? He (01:19:20) actually never won the finals (01:19:21) unfortunately for him. But the point is (01:19:23) like every soccer player dreams about (01:19:25) being in the Champions League finals, (01:19:26) right? Or a football player. Like the (01:19:29) pressure. Can you imagine the pressure (01:19:31) when you're going into that stadium? (01:19:32) Everyone's screaming. This is the kind (01:19:34) of, you know, the the height of your (01:19:36) career. This is the chance you got at (01:19:38) winning this thing. Like that's what I (01:19:41) signed up for. I signed up for being on (01:19:43) that interview with MSBC. This is what I (01:19:45) signed up for. It is stressful. It was (01:19:48) hard as hell, but this is what I wanted. (01:19:50) I wanted to play in this level of (01:19:52) league. So, I also have to cherish and (01:19:55) and and be happy about the fact and and (01:19:57) look at it and be great with gratitude (01:20:00) that I get the fantastic opportunity in (01:20:02) life to experience these things. It's (01:20:04) amazing. I'm going through this amazing (01:20:07) experience and that's the only way to (01:20:08) look at it. So as much as it half and I (01:20:10) can cry and I cried and I can be sad and (01:20:13) you know I've had really tough times (01:20:15) where I feel like super depressed about (01:20:17) stuff but at the same point of time I'm (01:20:20) always always looking at it putting on (01:20:21) under pressure and being like but this (01:20:23) is what I signed up for. (01:20:24) >> Yeah. I I love that too. (01:20:26) >> Yeah. (01:20:26) >> Did you ever interview him? (01:20:27) >> Uh do you know what? (01:20:28) >> You should bring him here. (01:20:29) >> No, we should do. And there's a (01:20:30) brilliant like video where he's like (01:20:32) when I step on the pitch I think I am (01:20:34) God. (01:20:34) >> Yeah. I don't think I'm God though. But (01:20:37) yeah. Do you did you always think you (01:20:39) would succeed there? (01:20:40) >> No, but it's exactly same. There's funny (01:20:42) there's this so there's this email that (01:20:44) I found that I wrote only six months (01:20:47) into the company and I found this email (01:20:49) just by coincidence and it's to my (01:20:51) co-founders, right? And it basically (01:20:53) goes like this. This is like six months (01:20:54) into the company. We just started. We (01:20:56) getting our first customers and then the (01:20:58) and basically the email goes like this. (01:21:00) I'm sitting here. It's late evening. (01:21:02) It's written like 11:30 p.m. you know (01:21:04) something. Yeah. And he goes, I'm (01:21:06) sitting here myself and I started (01:21:08) thinking like, you know what? What if (01:21:10) like we're actually successful with this (01:21:11) thing? Like what if this thing like what (01:21:14) if we start like growing maybe from (01:21:16) Sweden to Finland and Norway and then we (01:21:18) go to Germany? What if we actually grow (01:21:20) this globally? What if we actually go (01:21:22) and then we go after the banks and we (01:21:24) start building financial services? And (01:21:26) so I basically in that email I write (01:21:28) everything that's happened the last 20 (01:21:30) years. So the point is that obviously (01:21:32) did I know that was going to happen? No, (01:21:35) but just like Slatan when he was kicking (01:21:37) a ball down in Malner, you know, like on (01:21:40) the street, did he dream about being at (01:21:41) the finals? Of course he did. And of (01:21:43) course I did as well. I dreamt about (01:21:46) being where I am now and even more so (01:21:48) where I want to take the company in the (01:21:50) next decades. Right. So like (01:21:53) >> you dreamt about being here with me. (01:21:54) >> Of course. (01:21:55) >> Sweet Sam. I I did see it. (01:21:57) >> This was a very vivid dream. (01:22:00) >> But like I think visions are [ __ ] (01:22:02) All VCs are say, "Hey, what's your (01:22:04) vision, right, for a preede company?" (01:22:06) Dude, if I told you that when you were (01:22:07) starting your business, you know, 20 (01:22:09) years ago, you would not have been like, (01:22:10) "We're going to be a fullyfledged (01:22:12) banking provider. We're going to have (01:22:13) BNPL as the insertion point. We're going (01:22:15) to own the US as well." You would not (01:22:18) have been like that. You unlock the next (01:22:20) chapter with every achievement in my (01:22:22) mind. And so, I think visions are the (01:22:24) most actually constraining thing that we (01:22:26) force founders to try and articulate. Am (01:22:28) I wrong? (01:22:29) >> Well, I think you're wrong in in you're (01:22:32) wrong with me. I think I mean and and (01:22:34) again I don't I'm not telling you that I (01:22:36) know how we were going to do it. I had (01:22:38) no clue how uh but I was dead dedicated (01:22:43) to doing it and I did see or when I was (01:22:46) when I you know I I don't know why when (01:22:48) I was a kid I was an immigrant kid. My (01:22:50) my parents quarrel a lot. They divorced (01:22:53) when I was eight. My dad started (01:22:54) drinking. You know, it was quite a, you (01:22:58) know, chaotic upbringing and and in that (01:23:02) my childish interpretation (01:23:05) of what was happening in the family was (01:23:08) that they were always fighting about (01:23:10) money. If I fix money, then everything (01:23:13) will h end happily ever after. That was (01:23:16) my childish interpretation. And so I got (01:23:18) for whatever reason also very interested (01:23:20) in like Richard Branson and I read his (01:23:23) book and I now met him which was a (01:23:25) fantastic experience for me. But like I (01:23:27) read him and I was just like wow you (01:23:28) know this guy built Virgin did the (01:23:30) records and this and that. And then the (01:23:32) other guy that really inspired me was (01:23:34) Invarra the founder of IKEA you know who (01:23:36) built this at he was at for a period of (01:23:38) time was seen as the most wealthiest man (01:23:40) in the world right and so like I just (01:23:43) for whatever reason I was always like so (01:23:45) enthusiastic about building businesses. (01:23:46) I remember we we had this like school (01:23:49) night in in middle grade where like the (01:23:51) cafeteria was always the place that did (01:23:53) the most money and then there was like (01:23:55) tons of other businesses but they all (01:23:56) failed and nobody was really making a (01:23:57) lot of money but the point was to raise (01:23:58) money to go to like on a class trip or (01:24:00) whatever and I started a pizzeria you (01:24:02) know I started selling pizzas and we (01:24:03) outco competed the cafeteria and we took (01:24:05) all the money and my class could go on a (01:24:07) nice school trip because we made all the (01:24:08) money that the cafeteria was making. Um (01:24:11) so like I've always wanted to like drive (01:24:13) a business. I've always I've always had (01:24:15) this and then eventually I I said to (01:24:17) myself the coolest business to build (01:24:19) must be a bank like that must be the (01:24:21) ultimate business cuz banks like banks (01:24:23) always wins banks always prevail like (01:24:26) banks you know they you see these things (01:24:28) you see like the JP Morgans of the world (01:24:30) and like all this and they like they go (01:24:31) through this yeah of course like (01:24:32) sometimes they get you know financial (01:24:34) crisis and whatever but like in the (01:24:36) longer term the bankers always wins you (01:24:38) know so that was always an inspiration (01:24:39) to me (01:24:40) >> can I ask a hard question I I've (01:24:41) invested in 13 unicorns Um, well done (01:24:44) me. VC's congratulating themselves. Uh, (01:24:47) and uh, the single most common feature (01:24:50) is a broken relationship with their (01:24:51) father. (01:24:52) >> Um, I too have that. My father too has a (01:24:55) drink problem or had a drink problem. (01:24:57) Whatever. Um, (01:25:00) >> did that drive you? (01:25:01) >> It's not whatever though. Did he quit? (01:25:03) >> No. (01:25:04) >> Okay. (01:25:05) >> Uh, I don't really see him. (01:25:06) >> Um, (01:25:07) >> that's too bad. I'm sad for you. (01:25:09) >> Did that drive you in a way that you (01:25:11) wouldn't have had otherwise? if you (01:25:12) don't mind me asking. (01:25:13) >> No, I think for sure I think it's a (01:25:15) combination of that and being an (01:25:16) immigrant kid. I think that the the uh (01:25:20) you know, seeing all these other Swedish (01:25:23) kids going to their summer holiday (01:25:25) houses and having you know a surplus (01:25:28) that to me looked great. And then us (01:25:32) eating pancakes, you know, seven days in (01:25:35) a row because it was the cheapest food (01:25:37) that mom could put together because we (01:25:39) were out of money, you know, even though (01:25:40) I love pancakes. and I thought they were (01:25:42) really delicious. Um, I think it created (01:25:45) this like I felt my my parents were (01:25:47) smart. My dad deserved something like (01:25:50) better is the wrong word, but like he (01:25:52) started driving a cab, you know, and he (01:25:54) was very smart, very intellectual. He (01:25:56) that was not the right job for him, you (01:25:59) know. He was saddened when he saw drunk (01:26:02) people and he was driving them and (01:26:04) seeing what people were saying. And you (01:26:06) know, I think he was just not his place (01:26:09) to be in life. And it just it broke him (01:26:11) down. And and I felt that like if I was (01:26:14) going to fix things, if I was going to (01:26:17) get money, then I was going to fix this. (01:26:18) Now, the problem was once I got money, (01:26:20) it turns out that life isn't like that (01:26:22) because I I gave a lot of money to my (01:26:24) father and he used it to drink more and (01:26:27) he drank himself to death. So like it it (01:26:30) turned out that money was not going to (01:26:32) solve those problems. There are some (01:26:33) problems that money won't solve, right? (01:26:35) And again, I'm not talk I'm not I don't (01:26:36) want to have that discussion about like, (01:26:37) you know, does money make you happy or (01:26:39) not? Blah blah blah. Because obviously (01:26:40) when you have when you've been as (01:26:42) fortunate as I have in life and I can (01:26:44) take my kids on an amazing vacation, I (01:26:45) can do things and I don't have to think (01:26:48) every day about, you know, can I afford (01:26:49) this and can I do this? That's an (01:26:52) tremendous luxury and privilege in life (01:26:54) to be in that position. Um but at the (01:26:57) same point of time I've also experienced (01:26:58) that like it doesn't solve all problems (01:27:00) like and it's it's sad because that was (01:27:02) partially my aspiration with doing this. (01:27:04) Right. (01:27:05) >> Final one. I like to finish on a like (01:27:06) theme of positivity. What are you most (01:27:09) excited for in the next 10 years? Like (01:27:11) my mother's got MS. I'm very excited for (01:27:13) developments with diseases and (01:27:14) treatments for diseases like MS. What (01:27:17) are you most excited for? Well, I am (01:27:19) look I I am I'm I'm just very very I I (01:27:24) think with AI obviously the thing is (01:27:26) that if you five 10 years ago could say (01:27:28) that like well I kind of think I know (01:27:31) what's going to happen in the future (01:27:33) like it's going to kind of continue like (01:27:34) this or whatever and then suddenly came (01:27:36) covid and the war in Russia and you know (01:27:38) those things like the world just changed (01:27:39) and suddenly cames AI and you suddenly (01:27:41) sit there and like I have no idea how (01:27:43) the world is going to be in two years (01:27:44) like I have no clue like um but but for (01:27:47) myself what I'm most excited about is (01:27:49) like I a I do I'm still an optimist at (01:27:52) heart. I do believe that these (01:27:54) technologies will make life better for (01:27:56) humans. I think it will actually lead to (01:27:58) something positive. I'm in that camp. (01:28:01) Not I, you know, you can have an (01:28:02) intellectual debate with me if you want (01:28:03) to, but like I think that's true. And (01:28:05) then what I'm most excited about for me (01:28:07) is I want to realize the vision of (01:28:10) Clana. Like against what you said about (01:28:12) not having visions, like I want to put (01:28:13) that vision into reality. I want to (01:28:15) bring finally a banking product that (01:28:18) truly helps people save time, save (01:28:20) money, be in control of their finances. (01:28:22) That excites me. It excites the hell out (01:28:24) of me. And I think that like all these (01:28:26) incumbents have been having all these (01:28:28) excess profits. They've made so much (01:28:29) money because people don't switch (01:28:31) because and and honestly because they (01:28:33) didn't care enough about their (01:28:34) customers. They didn't wake up like that (01:28:36) restaurant or retailer every day and (01:28:37) said, "What can I do to make my customer (01:28:39) better off?" They didn't do that. And (01:28:42) and I'm I'm honestly excited about that. (01:28:44) that that journey is exciting to me. And (01:28:46) the point is and I know that if I make (01:28:49) more money, I'm not going to be happier (01:28:51) because I have a bigger pile of money. (01:28:53) It's not the point. But the journey of (01:28:56) trying to accomplish to make Clana into (01:28:58) that global retail bank and the (01:29:01) adventure it encompasses that excites (01:29:03) me. going through all these different (01:29:05) challenges and opportunities and trying (01:29:08) to make the best that I can of of (01:29:11) delivering on that that that thing just (01:29:13) really excites me. And now AI is (01:29:15) enabling me to do things that I couldn't (01:29:18) do with this company before. I can (01:29:19) realize those visions faster and at a (01:29:22) higher quality uh than was ever possible (01:29:24) before. And that's super exciting. (01:29:26) >> Dude, I've so enjoyed this. It's Do you (01:29:28) see it's so much nicer to do in person. (01:29:30) Thank you so much for being so brilliant (01:29:31) and being so open, but it's been such a (01:29:33) joy. Thank you. It's been a joy to be (01:29:35) here.