Building Lovable: $10M ARR in 60 days with 15 people | Anton Osika (CEO and co-founder)

0:00 lovable is your personal AI software 0:02 engineer you describe an idea and then 0:04 you get a fully working product the 0:06 reason is to enable those who have had 0:10 like such a hard time finding people who 0:12 are good at grading software that's been 0:14 their absolute bottleneck and let them 0:16 take their ideas and their dreams into 0:18 reality you guys hit 4 million AR in the 0:21 first 4 weeks you hit 10 million AR in 0:24 the first 2 months with just 15 people 0:26 you're the fastest growing startup in 0:28 all of Europe how did you decide on 0:29 lovable as the name it's so sweet the 0:32 best word for a great product is that 0:33 it's lovable a lot of jargon that I like 0:35 to use to like emphasize what we should 0:38 be striving for is building a minimum 0:39 lovable product and then building a 0:41 lovable product and then building an 0:43 absolutely lovable product so I I took 0:45 that jargon with me in the company name 0:47 people wonder just what jobs will be 0:49 more important what skills will be less 0:50 important doing a bit of everything 0:52 being in generalist is I think much more 0:54 important than it used to be if I'm 0:56 putting together a product team today I 0:58 I would really obsess about getting as 1:00 many skill sets as possible for each 1:02 person I hire what have you done that 1:04 has allowed you to grow this fast with 1:06 so few people people love the product 1:09 that's the driver of the 1:14 growth today my guest is Anton OC Anton 1:18 is co-founder and CEO of lavable which 1:21 is essentially an AI engineer that takes 1:23 an English prompt and codes a product 1:25 for you in minutes you can then talk to 1:27 it iterate on the product and then 1:28 launch it to the world it's one of the 1:31 fastest growing products in history the 1:34 fastest growing startup in Europe ever 1:37 and as Anton describes their goal for 1:39 lovable is for it to be the last piece 1:40 of software that anybody has to write 1:43 because it'll be able to create all 1:45 future products for us they launched 1:47 just a few months ago in the first 4 1:49 weeks hit 4 million ARR in the first 2 1:52 months crossed 10 million ARR all with 1:55 just 15 people absurd in our 1:58 conversation we covered a lot of 2:00 including a live demo of lovable how 2:02 their team operates how they hire what 2:04 do most enable their team to scale this 2:06 quickly with so few people Pro tips for 2:08 using lovable how it all started how he 2:10 recommends you build product TPS going 2:12 forward with tools like this existing 2:14 what skills will matter more and less 2:16 going forward plus how to think about 2:18 lovable versus competitors and so much 2:21 more if you're trying to wrap your head 2:22 around how product building will change 2:24 with the rise of AI tools this episode 2:26 is a mustat if you enjoy this podcast 2:29 don't forget to subscribe and followed 2:30 in your favorite podcasting app or 2:32 YouTube also if you become a yearly 2:34 subscriber of my newsletter you now get 2:37 a yearfree of perplexity and notion and 2:40 superhuman and linear and granola check 2:43 it out at Lenny 2:44 newsletter. with that I bring you Anton 2:48 OC this episode is brought to you by 2:51 cinch the customer Communications Cloud 2:54 here's the thing about Digital customer 2:56 Communications whether you're sending 2:57 marketing campaigns verification codes 3:00 or account alerts you need them to reach 3:02 users reliably that's where cinch comes 3:04 in over 3:06 150,000 businesses including eight of 3:08 the top 10 largest tech companies 3:10 globally use ca's API to build messaging 3:13 email and calling into their products 3:15 and there's something big happening in 3:17 messaging that product teams need to 3:18 know about rich communication services 3:21 or RCS think of RCs as SMS 2.0 instead 3:25 of getting text from a random number 3:27 your users will see your verified 3:29 company name name and logo without 3:30 needing to download anything new it's a 3:33 more secure and branded experience plus 3:35 you get features like interactive 3:37 carousels and suggested replies and 3:39 here's why this matters US carriers are 3:41 starting to adopt RCS cinch is already 3:44 helping Major Brands send RCS messages 3:47 around the world and they're helping 3:48 Lenny's podcast listeners get registered 3:50 first before the Rush hits the US market 3:53 learn more a get started at cinch.com 3:56 Lenny that's s i nch.com 3:59 / 4:01 Lenny this episode is brought to you by 4:04 Persona the adaptable identity platform 4:06 that helps businesses fight fraud meet 4:08 compliance requirements and build trust 4:11 while you're listening to this right now 4:13 how do you know that you're really 4:14 listening to me Lenny these days it's 4:17 easier than ever for frauders to steal 4:19 pii faces and identities that's where 4:23 Persona comes in Persona helps leading 4:25 companies like LinkedIn Etsy and twilio 4:28 securely verify individual and 4:30 businesses across the world what sets 4:32 Persona part is its configurability 4:34 every company has different needs 4:36 depending on its industry use cases risk 4:39 tolerance and user demographics that's 4:42 why Persona offers flexible building 4:43 blocks that allow you to build tailored 4:45 collection and verification flows that 4:48 maximize conversion while minimizing 4:50 risk plus persona's orchestration tools 4:52 automate your identity process so that 4:54 you can fight rapidly shifting fraud and 4:56 meet new waves of Regulation whether 4:59 you're star up or an Enterprise business 5:01 Persona has a plan for you learn more at 5:04 withp persona.com Lenny again that's 5:07 with P RS o n 5:10 a.com 5:14 Lenny Anton thank you so much for being 5:17 here welcome to the podcast it it's a 5:20 pleasure to talk to you Lenny great to 5:22 be here I don't know how you have time 5:23 to do this podcast your life must be 5:25 insane these days with the uh the pace 5:28 at which you guys are scaling just how 5:30 much is changing in AI every day uh so I 5:32 just extra appreciate you making time 5:34 for this I think you said it's 10:30 5:37 your time is when we're doing this I'm a 5:39 bit tired yes mostly from the the crazy 5:43 pace of everything but yes we're gonna 5:45 this is gonna be a invigorating 5:46 conversation you're not gonna be able to 5:48 sleep I'm sure I'm sure okay so for 5:52 folks that are maybe a little bit 5:53 familiar with lovable or not at all 5:55 familiar what's just what is lovable 5:57 what's the simplest way to understand it 6:00 I'd say lovable is your personal AI 6:02 software engineer you describe an idea 6:05 and then you get a fully working product 6:08 that from the and what this means is 6:11 that entrepreneurs actually today they 6:14 turn their ideas into real businesses 6:17 and we have a lot of designers and 6:19 product managers that uh create the 6:21 first version of of their product ideas 6:23 to show to their teams and and some of 6:26 them become Founders because of like 6:28 their the empowerment from 6:30 um but also developers themselves they 6:33 actually writing code or creating 6:35 products much faster and um I mean the 6:39 the reason it's pretty obvious for me so 6:42 I'll spell it but I'll spell it out the 6:44 reason why we're doing lovable is that I 6:48 don't know about your mom but like my 6:50 mom doesn't write code and my 6:54 friends almost all my friends from 6:56 throughout my life reached out for help 6:58 like Anon I want I do I need to build 7:00 something how do I find a great software 7:02 engineer and we're building for this 99% 7:06 of the population who don't write write 7:08 code um currently if you're technically 7:11 inclined you get much further but over 7:14 time 7:15 naturally the way to build software is 7:18 by just talking to an AI That's How is 7:20 it I love the way you guys describe it 7:22 and uh you didn't mention but I think 7:24 it's like building the last piece of 7:25 software ever how do how do you how do 7:27 you phrase that yeah we say we say we're 7:29 building the last piece of software the 7:31 last piece of software okay we're going 7:33 to do a live demo but first of all can 7:35 you just share some stats on the scale 7:38 of this business at this point because 7:40 it's quite 7:41 absurd yeah so we launched lavable less 7:45 than three months ago and now we have 7:48 300,000 monthly active users and 30 of 7:51 those 30,000 of those are actually uh 7:54 paying uh and the it is growing on this 7:56 at the same rates like you you just 7:59 almost only through organic word of mou 8:03 okay and uh I'll share a couple stats in 8:05 terms of Revenue just so folks know this 8:07 and we'll have this an intro too I think 8:08 you guys hit four million AR in the 8:11 first four weeks you hit 10 million AR 8:13 in the first two months with just 15 8:16 people you're the fastest growing 8:18 startup in all of Europe and you guys 8:21 had to rewrite your entire code base 8:23 recently and you couldn't ship any new 8:24 features for a while is there right 8:26 that's that's right yeah people were 8:27 saying like oh you're shipping so fast 8:29 and we were all quite frustrated because 8:32 we wrote our service in this you know 8:35 kind of scripting language and then as 8:38 we started scaling we we were just now 8:39 we have to throw everything away and 8:41 rewrite it in a more performant way okay 8:44 uh before we get to demo last question 8:46 you shared there's some companies that 8:47 have started based on lovable I didn't 8:49 even know that so what are some examples 8:51 of companies businesses that have 8:53 launched off of lovable and now are 8:55 actually companies I I mentioned the 8:57 signers using lovable and and one of our 8:59 early users Harry he he started 9:03 shipping real web apps to his clients 9:05 instead of just shipping the signs and 9:08 then he went on to say okay wait I'm 9:09 going to start an AI startup and and his 9:12 he his company he like launched on 9:14 product hunt and everything and making 9:16 uh money is just like lets anyone upload 9:19 their photo libraries and then it's C 9:22 like Ai pares and C categorizes it and 9:25 if you go to launched. lovable. apppp 9:28 like this is an app build Bel lovable 9:30 which is again a product product hunt 9:32 version where you can see a lot of 9:34 businesses or small s featured there 9:38 okay cool so we're g to come back to 9:39 some of the stuff but let's get into a 9:41 let's get into demo I I rarely do demos 9:44 on this podcast but I'm finding that uh 9:46 I think it's really important for people 9:48 to see these products in action because 9:51 in the large part this is the future of 9:53 product building and a lot of people 9:55 hear about oh yeah AI is coming and I 9:57 don't think a lot of people actually see 9:58 what the latest tool are capable of and 10:01 so uh I love showing these sorts of 10:03 things on this podcast H so Lenny I was 10:06 thinking um did you ever consider making 10:08 a copy and build your own adbn this I I 10:15 haven't but go on how about you do that 10:18 let's do it let's do it okay so we're 10:19 gonna make our own Airbnb okay so I I 10:22 just put in the first prompt for an rbnb 10:26 clone okay let's look and what what the 10:29 prompt just to for folks that aren't 10:30 watching two words rbnb clone that's the 10:33 pr I like she start INF and then what 10:36 you get is that the AI says okay I'm 10:40 going to go through what does a 10:42 beautiful rbnb clone look like and it 10:45 goes a bit of like decision design 10:47 decisions and then I'll zoom out to see 10:49 more of it h we we have this just UI 10:54 that is I mean it has all the the nice 10:56 things you would expect from araban 10:58 mclone where 11:00 um you see different categories and you 11:02 can see two listings fromb with login 11:06 buttons and everything so far it doesn't 11:09 have the functionality of rbnb it just 11:11 has the UI I would now ask for an 11:14 improvement on some of the functionality 11:17 like if I'm switching category I want to 11:19 see different listings let's say but if 11:22 you if you have any thoughts on what we 11:23 should build next let me know okay and 11:26 so you have this pre-loaded so you don't 11:27 see how long it would take but how long 11:28 would this normal take for it to just 11:30 write all this code and have it for you 11:31 the the first prom takes 30 seconds 30 11:34 seconds okay and it's like a very good 11:37 copy of Airbnb yeah I love that you 11:39 didn't have to show it to design you 11:41 just tell it Airbnb and it knows okay so 11:44 your question is what what I want to add 11:46 to my own version of Airbnb what I've 11:48 always wanted to explore buying the 11:51 place that I look at just like is this 11:53 for sale so what if we see what that 11:55 would feel like if you're just like a 11:57 way to buy buy a listing okay so let's 12:01 let's how about we add I mean prompting 12:04 is important here so let's be specific 12:06 but we would ask um creating a add a 12:10 button on the listing which say purchase 12:13 this this Airbnb home is that it perfect 12:17 so add I've 12:20 got and I'll be even more specific it 12:25 will pop up a model um to purchase the 12:31 listing perfect and I love so I think 12:34 some things as you're typing I'm just 12:35 going to share thoughts as you're doing 12:36 this so the site that you ask this AI 12:40 engineer to build like it's actually a 12:42 functioning website you can browse 12:44 around it's not just a design the say 12:47 obviously there's no like actual 12:48 listings here like there's no actual 12:50 houses here say you were trying to like 12:52 actually build Airbnb and you wanted to 12:55 start adding like actual homes that plug 12:57 into this how does that sort of Step 13:01 work so as you said this is just kind of 13:05 the mockup UI but it's Al also 13:08 interactive if I want to add login and 13:12 add listing 13:14 management 13:16 then we would connect something called 13:19 The backend so where data is stored 13:20 where users Lo information is stored and 13:23 I can show you how how to do that um 13:26 first let's just try out where we got 13:28 with this short prompt of adding the 13:31 adding the purchase uh listing and it it 13:34 didn't do exactly what I wanted I said 13:37 uh add um a button or I didn't say what 13:40 a button should say but it says book now 13:42 and if I click book now I get uh the a 13:47 booking confirmation so the the AI was 13:50 like okay it didn't really it was 13:52 probably surprised by you wanting to buy 13:55 the listing since it's rbnb right so it 13:57 still says book the listing but it's 13:59 shows a a pretty model where I can click 14:02 confirm and pay and then it says yeah 14:04 booking confirmed I'll just say real 14:06 quick I love that this is actually a 14:07 really good example of why being a good 14:10 product manager is important uh a lot of 14:13 wasted time happens when you're not 14:15 clear about the problem you're trying to 14:16 solve and why you're trying to solve it 14:18 and all that kind of stuff so it's 14:20 really cool that this is a use case 14:21 where you have to be really good at 14:23 explaining what it is you want and it's 14:25 interesting you don't have to tell this 14:27 a this aiy you know humans want to 14:29 understand why is this important uh 14:31 mostly you need to be very clear about 14:33 what it is you're doing uh and I love 14:34 that's a really strong PM skill you know 14:36 PM's really good at that so we have that 14:39 explaining exactly what you expect and 14:42 what you're not getting is even more 14:44 important with AI than with the humans 14:47 but um so so I go into hooking up more 14:51 of the actual functionality but first 14:54 I'll actually show you something um that 14:56 like how what's the fastest way to 14:58 change what went wrong it's it it 15:01 created uh buttons that say book now and 15:03 I want them to say um buy now and uh 15:08 what I could do is Select this item and 15:10 say change it to buy now but what we 15:14 just released is that you can actually 15:16 edit this like this is a fully 15:18 functioning product but you can edit it 15:21 visually like you can like you do in 15:23 square space and Wicks and so on so I'll 15:25 just change the text to BU now and then 15:28 it instantly it change just uh it 15:30 actually changes like deep down in the 15:32 code base but it's very fast to to do 15:35 that so I think people listening to this 15:37 and seeing this if you're not aware like 15:40 this is The Cutting Edge of tools like 15:42 this no other tool out there lets you 15:45 generate code from an AI engineer and 15:48 then actually just like change a small 15:50 element of it of every other tool that 15:52 I'm aware you have to like ask the agent 15:55 do this for me and then you hope that it 15:56 does the right thing so this is a huge 15:57 deal which you just showed 15:59 right yeah now it says by 16:01 now that's and that's something that you 16:03 just launched correct we just launched 16:06 it a few days ago but uh I want to go 16:08 into for building the full functionality 16:10 but what it looks like is that you 16:12 connect um an open source backend as a 16:16 service and that's called super base uh 16:19 and I have this instance to connect to 16:22 that's completely empty just like one 16:24 click to set that up and now it's 16:26 connected to uh the back end it's just 16:29 like automatically generating and 16:31 explaining um generating sound code and 16:33 explaining what I can do next and what I 16:36 would do now is say let's let's add 16:38 login let's say let's add login and 16:41 where is it actually hosted on the back 16:43 end and everything in general yeah 16:46 so everything can be one clicked 16:49 deployed and then it's running it's 16:51 hosted by like a cloud vendor which is 16:54 hosting I think a huge chunk of the 16:56 Internet it's called Cloud flare um and 16:59 the back end is hosted by the also good 17:03 cloud provider which is called super 17:06 based amazing okay uh let's wrap up the 17:10 demo that was unless there's anything 17:11 else was there anything else really 17:12 important you wanted to show no I mean 17:15 I'll just explain what you what I would 17:16 do next I would say okay let's add login 17:19 um let's make the listings editable by 17:22 the users so users can upload listings 17:24 and um then this is going to take a bit 17:27 more time but with patience and um good 17:30 prompting skills you're going to get to 17:32 a full working airbn that was a really 17:35 good uh piece to add so basically like 17:37 this is getting to a place where it 17:39 actually is not so different from actual 17:41 Airbnb uh people can log in they can add 17:44 their home you can add internal tools to 17:47 add listings for your say sales team Ops 17:49 Team basically it just will allow you to 17:52 build a market place that looks a lot 17:54 like careb um amazing okay thank you for 17:58 the demo I think for a lot of people 18:00 they're like yeah I've seen this kind of 18:01 stuff for most people like holy 18:04 it's unreal what like it's almost like 18:06 we're taking for granted now you can ask 18:09 a an app to build you a whole website 18:13 and that cost probably like a few 18:14 pennies it took like five minutes versus 18:18 like it would have been tens of 18:19 thousands in like weeks and weeks and 18:21 months Even build just a 18:22 prototype mean these tools as we see 18:25 here they're already very good like it 18:27 looks really good as well um but mainly 18:30 I I would say they getting uh better 18:33 very very fast and I I'd say like one of 18:36 the bigger bottlenecks is now they're 18:38 not integrated into the current way that 18:41 you have your existing products and so 18:43 on but since they getting better so fast 18:46 so fast I think the best thing for 18:49 people who are interested in this or 18:52 like interested in just being a part of 18:53 the future economies get your hands very 18:56 dirty with these tools because being in 18:58 the top 10% in using them is going to be 19:01 to absolutely set you apart in the 19:02 coming months and years so let's let me 19:05 follow that thread so say you are 19:08 magically able to sit next to everybody 19:12 that is uh using lovable for the first 19:14 time and you could just whisper a tip in 19:15 their ear to be successful with lovable 19:19 what would that tip be it it takes a lot 19:20 to master using tools like lovable and 19:24 being very curious and patient and I we 19:27 have something called chat mode where 19:28 you can just ask and like to understand 19:31 like how does this work like is I'm not 19:32 getting what I'm what I want here um am 19:36 I missing something what should I do is 19:39 is is the best way to be productive is 19:42 also one of the best ways to just learn 19:45 about how software engineering Works 19:47 which is you don't have to write the 19:48 code anymore but it is useful to 19:50 understand how software or how Building 19:52 Products works so that so I think that's 19:55 the patience and curiosity is super 19:58 useful the second part that we spoke 20:01 about is that 20:03 being I would if I Would S sit next to 20:05 you I would probably say like hey you 20:08 you're not being super clear here like 20:10 for example don't say it doesn't work 20:12 just explain exactly what you're 20:15 expecting and which parts are working 20:17 and which parts are not working and 20:19 that's a lot of that's something that a 20:21 lot of people don't do naturally I love 20:25 that like when you have an engineer 20:26 you're working with that is a very 20:28 expensive 20:29 uh mistake to miscommunicate something 20:31 to just forget about a future to forget 20:33 about a requirement and here it's you do 20:35 that and then like 30 seconds later 20:37 you're like oh okay sorry that was wrong 20:39 and then you could just try again that's 20:41 true it might it might be more costly 20:43 with 20:44 humans uh okay and the first step so the 20:47 first tip is chat mode so you could just 20:48 so your advice is chat with the what do 20:51 you call it do you call an agent you 20:52 call what's like the the term for the 20:54 thing that you were talking 20:55 with uh yeah lovable is just lovable 20:59 okay so you're talking about lovable by 21:01 the way where did you how did you decide 21:03 on lovable is the name it's so sweet I 21:06 think it's all about building I mean a 21:10 great product um that's what I I want 21:13 more people to be able to do and the 21:16 best word for a great product is that 21:18 it's 21:19 lovable the a lot of jargon that I like 21:22 to use to like emphasize what we should 21:24 be striving for is building a minimum 21:26 lovable product and then building 21:28 lovable product and they will an 21:30 absolutely lovable product so I I took 21:33 that jargon with me in the company name 21:36 that is great absolute level product Alp 21:39 the is the new MVP okay so we talked 21:43 about this the scale you guys have hit 21:45 at this point I imagine it's far beyond 21:46 10 million AR do you share that at this 21:48 point or you keeping that private we we 21:51 we don't an on numbers but I I could 21:54 probably do a twx tweet about this quite 21:56 soon yes okay so it's far beyond 10 21:59 million ER at this point uh it's uh one 22:02 of the fastest growing startups in 22:04 history the fastest growing startup in 22:05 Europe I want to zoom us back to the 22:07 beginning what is the origin story of 22:10 loable how did it all begin what was the 22:12 journey to 22:13 today I I think I was not impressed by 22:17 like what people were doing with the 22:19 large language models when after 22:21 especially after I was using them way 22:23 back but when CH gbt came out they were 22:26 starting to get really good at taking a 22:27 human instruction and spitting out code 22:30 and 22:31 then um people in my team I was the CTO 22:35 at a YC startup they felt like oh Anon 22:38 you're exaggerating this is not going to 22:41 change anything in the coming years so I 22:42 wanted to prove a point and I created a 22:46 open source tool called GPT 22:49 engineer where you you write something 22:52 like create a snake game and then it 22:54 spits out a lot of code a lot of 22:56 different files and then opens the snake 22:58 game 22:59 and then I tweeted a video about that 23:02 and um gbt engineer is today the most 23:06 popular uh open source tool to showcase 23:11 the ability for large language models to 23:12 create applications is that like 50 23:15 something 50 something thousand GitHub 23:18 stores and like dozen of academic 23:20 references and I know that I'll just add 23:22 that it like GitHub shut you down 23:24 because I thought it was some kind of 23:25 attack that like how many stars you 23:28 getting many people are using it right 23:29 yeah so that was that that came later 23:31 that that's with lavable thisable 23:34 lavable 23:36 um earlier was always creating new 23:39 projects on GitHub when someone used 23:41 lovable and it was we asked them is it 23:44 fine like how what's the limits here 23:46 they said oh there no limits but once we 23:48 started creating 15,000 projects per day 23:52 uh so there were a lot of usage then 23:55 some engineer when was on call maybe 23:59 they woke up in the night and they saw 24:01 their servers are were taking too much 24:03 load because of us so then they they 24:06 shut off down completely and we got this 24:08 email that said oh you broke some kind 24:10 of rules and we didn't know what was 24:12 going on that's some much just story I 24:14 heard when uh chat GPT was originally 24:16 being trained Microsoft servers were uh 24:20 blocked it because they thought it was 24:22 some crawler and it was just actually 24:23 like the very first version chat gbt 24:25 being trained on on data anyway keep 24:28 going 24:28 and so I I built this tool called GPT 24:32 engineer and um I was thinking about I 24:37 mean we're go we're seeing the biggest 24:40 change Humanity will ever see I think 24:42 where like before you had manual labor 24:45 being taken over by by machines but now 24:47 it's actually cognitive labor being T 24:49 being done better than humans by 24:52 machines and what's the best way to have 24:55 some kind of positive impact here it's 24:58 not to make Engineers more productive 25:00 which is there's a lot of companies 25:01 using a to make Engineers more 25:03 productive Microsoft you build co-pilot 25:04 and so on but it is to enable those who 25:09 have like such a hard time finding 25:12 people who are good at creating software 25:13 that's been their absolute bottleneck 25:16 and let them take their ideas and their 25:19 dreams into reality so enabling more 25:22 entrepreneurship and The Innovation by 25:25 building the AI software engineer for 25:26 for anyone and then I I put I grabbed a 25:30 previous colleague of mine who has also 25:32 been a Founder Fabian and I said we we 25:34 should build something like GPT engineer 25:37 but it it has to be for the people who 25:39 don't write 25:41 code that's the story okay and then that 25:44 became lovable there's like the shift 25:46 from open source into a product that 25:48 anyone can use but also pay for makes 25:51 sense okay so from that point uh I saw 25:54 stat they started making a million 25:56 dollars in AR per week and once you 25:58 launch lavable is that true yeah so we 26:01 launched um so we actually called the 26:04 first version of the product like GPT 26:06 engineer app uh and that that's it was 26:09 very different in some ways um and we we 26:12 launched that under a weit list and so 26:14 like oh yeah we have this weight list 26:16 and we got a lot of feedback and 26:17 iterated and finally when we thought the 26:20 product was really good we said okay now 26:21 we have a lovable product and it was 26:24 mainly on the AI that we did a lot of 26:26 improvements 26:28 once we launched that that was 21st of 26:30 November so that's almost three months 26:32 ago we uh just hit like one mill million 26:36 AR in a week and then it kept go growing 26:38 at that that pace it's still growing at 26:41 even faster than that pace faster than 26:43 one million AR per week holy okay 26:48 that sounds like product Market fit to 26:49 me you said that you did a lot of work 26:52 on the back in I saw you tweet about 26:53 this that you guys figured out some kind 26:55 of unlock on scalability like a new SC 26:58 scaling law that allowed you to build 27:00 something like this what can you talk 27:02 about there that kind of on the 27:03 technical element allowed you to build 27:05 something new and and the successful 27:08 there are many scaling laws I would say 27:09 when you build AI systems and this one 27:12 in particular is about when you put in 27:14 more work this the pro the product 27:16 reliably gets better and better and what 27:20 you what you've seen um generally when 27:24 you have ai building something is that 27:26 it can get stuck in in some place it 27:28 stars is super good in the beginning and 27:30 then it gets stck what we did was to 27:33 painstakingly identify places where it 27:36 go stuck and um there there's a 27:39 different approaches but address like 27:42 different ways how we do it but address 27:44 the places where it gets like tuned the 27:46 entire system quantitatively and having 27:50 a very fast feedback loop to improve it 27:52 in the areas where it got stuck the most 27:54 important areas it still does get stuck 27:56 sometimes but that's the scale law and 28:00 um we're still early in that scaling law 28:02 I would say and so when you talk about 28:05 things getting stuck it's like the the 28:07 AI agent just saying like I don't know 28:08 what to do from this point and or like 28:10 they introduce some kind of bug is that 28:11 is that an example of getting stuck 28:13 introduces some kind of bug and then um 28:17 it's not smart enough to figure out how 28:19 to get out of that bug I see and this is 28:21 a common a common problem people have 28:23 with tools like this is they like get to 28:25 a certain point and then it's like well 28:27 I don't know what to do I'm not an 28:28 engineer like here's a bug it's running 28:30 into or the infrastructure is built the 28:32 wrong way and so it sounds like uh one 28:35 of the paths to solving that is what 28:37 you're describing is you make the AI 28:40 smarter to get to avoid more and more of 28:42 these places they get 28:43 stuck another is people just learning 28:46 how to get AI 28:49 unstuck uh this something when we had 28:51 omj on the podcast from replate he said 28:52 that this is like the main skill that he 28:54 thinks people need to learn is how to 28:56 unstuck AI when it runs into a problem 28:59 uh just thoughts there I don't know 29:01 anything along those lines come up as I 29:02 say that me this is something that uh is 29:05 a problem today and it the frontier of 29:10 where this is a problem is very rapidly 29:14 like receding back so uh what we did was 29:18 we identify the most important areas 29:20 like oh so specifically adding login 29:23 creating data persistence 29:26 adding payment with stripe like those 29:29 are the things that we made sure it 29:30 doesn't get stuck on for example um and 29:33 the places where it gets stuck today um 29:37 is currently the something that where 29:39 you can use being very good at 29:41 understanding and getting unstuck but in 29:43 the future it won't be so important this 29:45 is just going to not get stuck and I 29:48 know you're you're not talking super in 29:50 depth about this because this is one of 29:51 your unfair advantages this kind of 29:53 stuff you figured out so I'm not going 29:54 to push too far I don't know I know you 29:56 want not everyone to do exactly the same 29:57 stuff stuff so I want to zoom back to 30:01 the pace of growth that you guys have 30:03 seen one of the big stories everyone's 30:05 always looking at you guys of like 15 30:07 people 10 million AR in two months 30:10 that's absurd it's something I don't 30:12 know if it's ever been done in history 30:14 if if so it's maybe a couple other AI 30:15 startups recently how have you been able 30:18 to do this what have you done that has 30:20 allowed you to grow this fast with so 30:23 few people I'd like to take credit of 30:26 like having done everything end to end 30:29 in the product um but what but we're 30:33 building on top of the kind of the oil 30:36 here which we have discovered oil which 30:38 is are the foundation models right um 30:41 and then what you what we've done is 30:42 that we obsessed about what's the right 30:44 way to present this to a user what's the 30:46 interface for the human to get as much 30:48 out of this as possible 30:51 packaging together I I showed you in the 30:53 demo that you how you can add 30:56 authentication and making this work 30:58 seamlessly together as a whole that 31:00 that's what we've done and then people 31:03 love the product that's what that's the 31:05 driver of the growth uh the for getting 31:10 awareness we' mainly been posting what 31:13 we've shipped on social media that's 31:15 that's how people know about us so 31:17 building in public is is is how people 31:19 usually describe that so it's like uh I 31:22 think it's like you guys have the 31:23 advantage of the demos are just like 31:24 holy you can do that and then you 31:26 guys share the numbers that you guys are 31:27 growing at so it's innately interesting 31:30 and 31:30 sharable uh but I imagine most people 31:33 have something interesting to share I 31:35 guess is there anything that you think 31:36 you did that other companies maybe 31:38 haven't done that make the product so uh 31:42 lovable that I mean the the te is 31:45 everything in building a great product 31:47 so I I just give a give a big shout out 31:50 to to te that has written the code I I 31:53 Haven written the code Rec much of the 31:55 code recently I would say um 31:58 and the I mean you you want people who 32:03 can ship really fast and have have good 32:05 taste for like what is simple what's the 32:08 right 32:09 abstractions and I think that's what 32:11 we've done differently and have have 32:13 this Obsession for us making it better 32:15 and better and 32:16 better this episode is brought to you by 32:18 the fundrise flagship fund full 32:20 disclosure real estate investing is 32:23 boring prediction markets are exciting 32:26 meme coins are A Thrill Ride even the 32:28 stock market can swing wildly on a 32:29 headline hello deep seek but with real 32:32 estate investing there's no drama or 32:34 adrenaline or excuses to refresh your 32:36 portfolio every few minutes just Bland 32:39 and boring stuff like diversification 32:41 and dividends so you won't be surprised 32:43 to learn that the funrise flagship real 32:45 estate fund is a complete snoozefest the 32:48 fund holds $1.1 billion worth of 32:51 institutional calber real estate managed 32:53 by team of Pros focused on steadily 32:55 growing your net worth for decades to 32:57 come see 32:58 boring that's the point you can start 33:01 investing in minutes and with as little 33:02 as $10 by visiting fund rise.com Lenny 33:07 carefully consider the investment 33:08 objectives risks charges and expenses of 33:11 the fundrise flagship fund before 33:12 investing find this information and more 33:15 in the funds perspectus at fundrise.com 33:18 Flagship this is a paid 33:21 ad okay I want to come back to the team 33:23 because I know you have a lot of 33:24 thoughts there in terms of writing code 33:26 how much do you guys actually use AI to 33:28 write the code that is building lovable 33:30 like how does that work on your team we 33:32 have set up lovable so that we can 33:34 change lovable with itself we have done 33:37 that um since there's a lot of 33:41 like hyp specific things um in terms of 33:46 running a separate like we spin up a 33:49 dedicated computer for each user uh it's 33:53 doesn't do everything lavable doesn't do 33:55 everything so we use we like 33:58 the tools are for developers not for the 34:00 99% most of the most of the time and uh 34:04 everyone uses AI all the time in writing 34:07 code it's also a great course for 34:09 experimentations and other tools like 34:11 cursor and stuff like that like any 34:12 tools you can sh I think cursor is the 34:15 um the one that almost every everyone 34:17 uses in in the team yeah okay cool we I 34:20 did a survey recently in on tools that 34:23 my listeners and readers use in cursor 34:25 like 177% of all people that read my 34:27 newsletter use cursor already which is 34:30 absurd and you guys are in there too 34:33 okay so kind of along these lines 34:34 there's obviously other competitors and 34:36 companies in the space so everyone's 34:37 always wondering uh you bolt repet 34:40 cursor is a different kind of thing 34:42 what's the simplest way to understand 34:43 maybe how Lev B might be different from 34:46 say bolt and ret which I think are 34:48 probably the 34:49 closest the packaging for non-technical 34:52 people is what we what we aim for and um 34:56 I showed you in the demo that that you 34:58 can edit the text like you can change 35:01 the colors and so on instantly with 35:04 without having to go into like code 35:06 editor um and without having to wait 35:08 this about 30 seconds for the AI to to 35:10 the full change so uh that's the the big 35:14 way that we think about packaging it and 35:16 then for you know making sure that this 35:21 can be used as productively as possible 35:23 in a larger team uh something that's 35:25 different from I think the other other 35:27 tools is that it's it is synchronized 35:30 with GitHub and that means that you can 35:33 use cursor if you're or the people in 35:35 your team that that want to be more low 35:38 level they can use cursor and while the 35:40 people who don't want to mess and set up 35:42 their local file system and commit to 35:45 GitHub and so so on they you can use 35:47 slbl not getting stuck is I I think the 35:50 most important thing for people and 35:51 that's why we came we can't enter the 35:54 space late we haven't done the same type 35:56 of marketing as many others and we're 35:58 still um from the people that I talk to 36:02 ranked as the one that works most 36:04 reliably I I love it okay so uh so this 36:08 point about how you can just use lava B 36:11 to build a lot of it for you and then 36:13 get into cursor to edit and tweak is is 36:16 a really big point and you're saying 36:17 other other companies aren't as good at 36:19 that yeah I don't know any other dust 36:22 that oh he W let you do that amazing 36:24 okay and then what's kind of like the 36:26 vision for lovable like what's the end 36:30 state of this is this everybody can 36:31 build anything they want sort of thing 36:33 what's the simplest way to understand 36:34 where you're going in the next I don't 36:35 know five 10 years I mean I have to say 36:38 so we're building the last piece of 36:40 software and it is inherently very hard 36:42 to predict how the world looks like in 36:44 five years these days it's very hard and 36:47 but the last piece of software how I see 36:49 that is that it's almost instant to go 36:53 from what you want to change in a 36:55 product or what you what product you 36:56 want to build to having it fully working 36:58 end to end integrated with any of your 37:01 existing systems or integrated with the 37:04 kind of the very powerful third party 37:06 providers already today you can just ask 37:09 add a chat with open Ai and then you get 37:12 a chat with open AI in your in your 37:14 product but um that's like just work 37:17 working perfectly is the something 37:20 that's coming in the coming two years I 37:22 would say um and then after that 37:28 there is a lot of things in building a 37:30 product that is not just engineering 37:31 side 37:32 right and I think um an AI can be very 37:38 useful in AGG aggregating and 37:41 understanding your users so like if you 37:44 have if you use the analytics tools you 37:46 know that there's something quite common 37:47 which is to see how users have 37:50 interacted with the product AI can do 37:52 that an absolutely massive scale and 37:55 propose changes to a human to say like 37:57 yeah that sounds like a good change to 37:58 make it a bit more intuitive and it can 38:01 also automatically run spin out AB tests 38:05 so that you can see with data all these 38:08 improvements to the product so that I 38:10 think that's on the horizon as well 38:12 quite 38:14 soon like what's interesting about this 38:16 in in one way is people wonder just what 38:19 jobs will be more important what skills 38:21 will be less important let me share a 38:22 thought I have and then I want to get 38:23 your take and see where you go with this 38:26 it feels like what is getting more 38:28 valuable as being good at figuring out 38:31 what to build and then knowing if the 38:33 thing you had built is correct and good 38:36 and ready so it's like Discovery 38:39 ideation uh 38:41 idea part of the step of launching a 38:43 product and then it's like taste and and 38:46 craft just like is this the thing is 38:48 this going to solve people's problems 38:50 because the the building now is being 38:52 done more and more and it's interesting 38:54 it used to be the reverse engineering 38:55 was the hardest most valuable skill and 38:57 now it's like figuring out what to build 38:59 you could sit there and you you could 39:00 just tell what to build and a lot of 39:02 people get to your screen I'm sure and 39:04 they're like I don't know what to build 39:05 I don't know what people want and it's 39:07 like that's the thing now so I just 39:09 reactions to that and thoughts on what 39:10 skills will matter more and less I mean 39:13 if you're if you want to if you're a 39:15 founder or you want to build something 39:16 yeah I I totally agree that figuring out 39:19 what are what are pain pain points and 39:22 seeing like there are pro often 39:25 currently solutions to every some kind 39:27 of solution to everything what is the 39:29 and how can you make this tenic better 39:31 somehow like figuring that out is super 39:34 important when you have um an existing 39:37 product then I think taste and like 39:39 refine tasting what is what is good is 39:41 even more of the important part the tech 39:46 like the engineer skill set is still 39:48 going to be important because that that 39:51 helps you understand what are the 39:52 constraints so what you can build and I 39:55 just think a lot of software Engineers 39:58 are probably a bit scared now like okay 40:00 am I out of a job what's going to happen 40:03 but they should see themselves as the 40:05 people who translate the the problems 40:07 that are stated by a human probably um 40:11 to Technical Solutions and but they do 40:15 have to abstract themselves up a few 40:16 steps not just like looking at the in 40:18 their text stack like oh I can just do 40:19 the front end changes they Engineers or 40:23 technical people are very good at 40:24 understanding what are the constraints 40:26 technically and they should see 40:28 themselves as that translators is there 40:30 like a like is it almost like you want 40:32 to be learn the enge manager skill of 40:35 overseeing Engineers versus like the 40:38 actual engineering skill or is you think 40:40 it's still going to be really important 40:41 to learn how to code and be really good 40:43 at that I mean doing a bit of everything 40:46 being in generalist is I think much more 40:48 important than it used to be and the if 40:52 if I'm putting together a product team 40:53 today I I would re obsess about getting 40:56 as much 40:57 of as many skill sets as possible for 41:01 each person I I hire like they should 41:03 know how architecting and system works 41:07 prably they should know design they 41:10 should know they should have product 41:11 taste they should know how to talk to 41:12 users I think everyone should be able to 41:14 know should know a bit of all of that 41:16 preferably easier said than done it's 41:19 hard to find people that know all these 41:20 things so let's Segway to hiring and how 41:23 you hire how many people do you have at 41:25 this point is that some you sure yeah 41:28 now we're at 18 18 okay wow so I love 41:32 that you it sounded like you're about to 41:34 say oh we have 100 people now no 18 okay 41:36 so you went from 15 to 18 uh okay great 41:40 so what do you look for when you're 41:42 hiring people the way saw you describe 41:43 it on Twitter as you look for cracked 41:45 Engineers the best cracked team in 41:48 Europe things like that I guess just 41:49 specifically what are you looking for 41:50 when you're hiring I think the most 41:52 important thing is that people care a 41:55 lot and they're not just like oh I'm 41:58 here for a job I'm here for being as as 42:01 a passenger on this journey but everyone 42:03 should really care about the product the 42:05 users and Care a ton about the team how 42:09 the team works together and that you're 42:11 always contributing to making the team 42:14 work more productively together and that 42:18 like care or preferably Obsession ER 42:21 gets you very long 42:23 way and then um you do 42:27 often want to have like absolute 42:29 absolute superpower of some Dimension to 42:32 be able to understand and do as many POS 42:34 things as possible like have this 42:36 generalist brain that quickly learns any 42:39 skills but be super super good in in one 42:42 dimension and that's for us that's of 42:44 that's mostly cramming as much out of AI 42:48 out of the large language models and 42:50 understanding the um the entire 42:53 parameter space of what you can change 42:55 to make this the our product to perform 42:57 better so how do you actually test for 42:59 these things you know like some of these 43:01 things described I think everyone's 43:02 looking for like they care about the 43:04 user they want to collaborate well just 43:06 like when you're because like you have 43:08 18 people building in a company that's 43:10 growing more than a million AR every 43:12 week like that's an absurd uh uh scale 43:16 and the people you found are clearly 43:18 world class and I think a lot of people 43:20 are going to like want to hire the type 43:22 of people you're hiring so when you're 43:24 actually interviewing how do you sus out 43:26 some of these things like like their AI 43:27 cramming skills their team building 43:30 collaboration what do you actually do I 43:33 ask people what they've done before and 43:35 they these people that I'm describing 43:38 they have often done something where 43:39 they care a lot about what they've done 43:41 before uh and dig into details about the 43:47 technical things that they did and then 43:50 um I mean we do the normal thing of 43:52 giving it showing a very hard problem 43:54 that is a bit unorthodox that someone 43:56 hasn't seen before preferably and see 43:58 how they think through the think and 43:59 reason through that I then something 44:02 that I think is more Uncommon is that we 44:05 do I pretty much always have people join 44:09 the work simulation for at least a day 44:11 often a full week awesome okay so work 44:14 trial that's awesome so basically they 44:17 work with the team for at least a day 44:19 you said like sometimes a week yeah and 44:22 uh I love this point you made about they 44:24 show they cared deeply about some they 44:27 previously worked on and you can you 44:29 look for just like obsession with the 44:31 thing that they built last or something 44:33 they worked on like what percentage are 44:36 Engineers of these 44:38 18 the 12 at least write code in at 44:44 least part times 12 out 18 okay cool uh 44:48 you when we were setting up you're like 44:49 oh our Engineers creating content now I 44:52 think that's a cool example of of how 44:54 people do a lot of different things yeah 44:56 uh also okay so I have your job posting 44:59 that you shared once of 45:00 like the actual job description I'm 45:03 going to read a few lines from it it's 45:04 uh very inspired by Shackleton right 45:07 would you agree cool I love it by the 45:09 way did you write this or did you have 45:11 ai write this job description where you 45:12 like create an engineering job 45:14 description FR let me read it to you I 45:15 don't even know you may not know 45:16 whatever you referring to uh I'll read a 45:19 few lines here long hours High Pace 45:22 candidates must Thrive under high 45:23 urgency under AGI timelines approaching 45:27 uh difficult Mission ahead honor and 45:29 recognition in case of success those 45:31 seeking comfortable work need not 45:33 apply and then there's a few other 45:35 things collaboration with other 45:36 exceptional Minds purpose larger than 45:38 any normal engineering role generous 45:40 share in the Venture success 45:43 amazing thank you thoughts yeah so I I 45:46 did I did get some help with the the 45:48 formatting of this but then I was mostly 45:52 me doing the the exact pring of the 45:54 different sentences so good and I love 45:57 that you know to some people it's going 45:59 to be like holy I'm not signing out 46:01 for this but to a lot of people the 46:02 people you want is like yes this is 46:04 exactly what I want to be 46:06 doing great amazing Okay cool so so it 46:10 feels like one of the elements of hiring 46:12 here is uh create a really good filter 46:15 to be clear about just how intense this 46:18 is so that the people that want that are 46:20 the ones drawn to 46:21 you okay and then you're also you're in 46:24 Sweden uh fastest growing startup in 46:27 Europe ever thoughts on building in 46:30 Europe Sweden versus the US San 46:33 Francisco yeah so this this ambition 46:35 level that you you're talking about in 46:38 the job is more Uncommon in Sweden and I 46:41 think that is the like the biggest 46:44 unlock that someone like me who sees 46:48 that this is the um like the time in 46:53 human history when you have the most 46:54 impact per worked hour and and that's 46:57 why we have to be super ambitious like 46:59 just up the ambition level and then then 47:00 we can maybe retire and have ai take 47:03 care of most most things in society um 47:07 that and inspiring people to be this 47:11 ambitious um in a place where the 47:15 average ambition is lower but the talent 47:17 the raw talent is um much more available 47:21 is is a great recipe I think that's a 47:23 great recipe so the and that's what 47:27 I think it's some kind of Advantage 47:28 there it's a bit of a double-edged sword 47:32 but but it's some kind of Advantage so 47:34 I'm hearing is like there's there's 47:36 incredible people in Europe they're just 47:38 not uh they're harder to find and what 47:41 I'm hearing is like the key is how do 47:43 you sus them out and get them to to want 47:46 to talk to you yeah most people in 47:50 Europe they haven't thought that oh do 47:53 going on an extremely ambitious mission 47:55 is what I want to do so that's that 47:57 figuring out who those are H is is a big 48:00 part of it awesome okay I want to talk 48:03 about priorization I 48:05 imagine all these things that I just 48:07 shared about just like how uh ambitious 48:09 this mission is how much you're doing 48:10 the last piece of software you must have 48:13 a bazillion things that people ask you 48:15 to build that you want to build what's 48:17 your approach to deciding what to biriz 48:19 and actually build I ju Just Top Line I 48:22 think identifying what is the biggest 48:26 ball neck what the biggest produ problem 48:28 and iterating fast on saying okay this 48:31 is the biggest problem let's really 48:32 really solve that that problem and then 48:34 picking the next one um and not 48:37 overthinking not like dreaming out the 48:38 long road map that's my my default it's 48:41 a very very simple algorithm um 48:44 understanding what is the most B the 48:46 biggest problem is not always a simple 48:48 simple problem I 48:50 think yeah so we spend time as one 48:53 should on talking to users list 48:57 reading up on what people are writing we 48:59 we have the F board for where people do 49:02 a lot of requests as you say and then um 49:07 when we pick one of the problems we're 49:10 quite engineering lead like for a 49:13 product like ours it's hard to be like 49:15 have uh product managers that are not 49:18 Engineers say oh this is what we should 49:20 do now 49:21 because like the right solution to the 49:25 problem might be in time Tangled in 49:27 things that are um like technical 49:31 details they might be entangled in 49:33 technical details of like okay yes this 49:36 is the biggest problem but we should Sol 49:38 we should have this larger technical 49:40 initiative that's going to solve all of 49:41 these problems so it's it's quite 49:43 engineering lead um compared to many 49:46 other product companies as it should I'd 49:48 be I'd be worried if you guys had a 49:50 product manager at this point makes that 49:52 would not that wouldn't make no sense 49:54 right now I imagine the answer is it's 49:56 chaos and there's no actual uh defined 49:59 process but just like what does it look 50:01 like generally like what's kind of the 50:03 Cadence you guys operate on how do you 50:04 take an idea to like build it spec it 50:07 launch it just like what does that look 50:09 like if you have something if you look 50:12 back like three months we mainly said 50:15 okay let's do this weekly planning and 50:18 we have we have like a fig Jam board 50:21 where we have all the main problems and 50:23 then we have kind of rank them which are 50:25 which on do we focus one more f on next 50:27 or this week um and then we have a demo 50:31 where we say like okay this are this the 50:33 things we Shi this week so they get 50:35 everyone on the same page and we do have 50:39 a bit more of a road map now and where 50:41 we say like here are we going to make so 50:45 sure you can support custom domains next 50:47 they're going to add 50:48 collaboration uh after that and and the 50:53 like the biggest problem now or the the 50:55 biggest initiative now that solves the 50:57 biggest problem is making the system 50:59 more identic um and that has a bit of a 51:02 longer road map but we still do the 51:03 kadence of weekly planning these are the 51:06 like the things we're focusing on this 51:09 week it's mostly there's a good word for 51:12 this that you I would want you help with 51:14 with Polish fixing the bags and and 51:18 polish this week and that was the 51:19 planning on Monday that was actually 51:21 this week was uh Polish Polish week I 51:24 love that uh how far is this road map 51:26 that you're now having I mean it's uh 51:30 clear over the coming month but it 51:33 stretches up three months and then but 51:35 with with in one month it's probably 51:37 going to look a bit different okay and 51:39 then what are the tools you use just for 51:40 folks that want to understand like the 51:41 latest tools so you said fig Jam what 51:44 else isn't that stack of tools I mean we 51:46 do so many things in our company in 51:49 linear because it's just such amazing 51:51 product so we we do Talent application 51:54 tracking in linear oh w 51:57 after going through and and this seeing 51:59 a l of the other made custom made tools 52:01 for that uh linear and then uh feed jum 52:06 so simple uh how soon until one of your 52:09 engineers is an agent engineer in AI 52:12 engineer do you think do you have a 52:14 sense I love to dig into what what does 52:17 that question actually mean um I think 52:20 we've been talking about like oh AI that 52:23 would require um or something playing 52:26 chess that's that's AI like if you if an 52:29 if a computer can play chess that's Ai 52:30 and now that's like oh no that's a chess 52:33 program and we always shifting this 52:36 forward and forward um I 52:39 think 52:42 anything that a human doesn't do is just 52:47 a smart computer system right so what 52:51 is when is some when is an a software 52:54 engineer and agent 52:57 I think it's always going to be just 52:59 we're building in lovable is just an 53:01 interface that humans interact with to 53:04 create the software that they want and 53:07 then how we solve that is that going to 53:09 be an agent under some definition yeah 53:11 sure I think so but uh that's less 53:14 important to me okay I like 53:17 that let me ask this you guys are moving 53:20 super fast scaling like crazy you 53:22 described a little bit about your 53:23 process weekly planning uh fig Jam Board 53:26 of is and now there's a road map that 53:28 you're kind of thinking out in the 53:29 future is there anything else that you 53:31 found was helps you move this fast that 53:34 gives you a lot of Leverage over the 53:35 small team you have to ship quickly and 53:37 move fast that you haven't already 53:39 mentioned we we work from the office 53:42 most of the time I think it's it's 53:44 pretty nice then you can say like hey I 53:46 think we're thinking wrong about this 53:48 thing or like shouldn't we actually do 53:49 this other thing and especially I think 53:52 lunch like eting lunch together is a 53:54 pretty productive hour 53:57 uh where you're cross-pollinating I mean 54:00 people are constantly thinking 54:02 subconsciously as well about the how to 54:04 solve the different problems and which 54:05 the most important ones are and then 54:07 being in office um has this like Focus 54:10 or most of the time you should be Focus 54:12 but you also have this like high 54:14 bandwidth where everyone has a bit 54:15 unstructured 54:17 communication I love that uh the answer 54:19 to uh the CEO of a company that's the 54:22 one of the most advanced AI tools in the 54:24 world is one of your answers to how to 54:26 move fast is like lunch together I love 54:29 that it's so human and so it makes all 54:32 the sense in the world but I love that 54:33 that's still a part of this 54:35 yeah okay you talked about this kind of 54:38 on the same thread you talked about if 54:41 you were to start in a team like a new 54:43 product team today say you were head of 54:45 product somewhere or head of RPM uh VP 54:48 of product somewhere building a new 54:49 product team scaling a product 54:51 team what would you do going forward 54:55 that's different 54:56 from what people have done in the past 54:59 in terms of who you're hiring how you're 55:02 structuring them that kind of thing just 55:04 like what do you think people should be 55:05 thinking as they build product teams 55:07 going forward knowing tools like lovable 55:09 exist and all the other stuff that's 55:11 going on I mean everyone should be 55:13 excited about using AI I think that's a 55:15 pretty big one um and then and the team 55:20 working really well together is is what 55:23 like the lunch you have to H like to sit 55:25 down and solve problems together um you 55:29 should at the bottleneck for most 55:34 products these days is not going to be 55:36 as much on engineering but having good 55:39 taste good intuition about your users 55:42 and um that I mean engineers and 55:47 everyone preferably in the team should 55:48 have that like willingness at least to 55:51 to want to go through that motion and 55:52 listen to the users um and 55:56 truly understand what what they care 55:59 about what's kind of like the background 56:01 of most of the engineers and people 56:03 you've hired are they like is there 56:05 anything like in common are they just 56:07 like 56:08 super uh impressive humans generally 56:11 like you know champions of programming 56:14 contest stuff like that I don't know 56:15 like what are some attributes of the 56:16 folks you've hired so far I think Ro 56:21 cognitive capability is the strongest 56:23 like di the the strongest correlate of 56:26 being at love lovable 56:29 but there there is this startup mindset 56:32 that I I think is also very strong being 56:35 a bit more being being much more 56:39 interested in moving very fast and uh 56:41 iterating fast than having like a lot of 56:44 structure a lot of 56:45 process and thinking about the business 56:48 as a whole more than thinking about my 56:50 specific profession my specific craft 56:52 that I'm see myself like wanting to dig 56:55 in into on me amazing okay so smart like 56:59 very smart entrepreneurial acts like an 57:03 owner yeah doesn't just uh isn't just 57:06 like this isn't just a job but they feel 57:07 like they actually have agency okay this 57:10 is great there's something you said kind 57:12 of along these lines that uh I think is 57:14 important that one of the things that 57:16 gets you excited about what you're 57:17 building is giving people superpowers 57:20 and especially people that don't out a 57:21 code basically 99% of people is there 57:24 anything along those lines that you 57:25 think is important to share it's very 57:28 clear to most people who have been 57:30 Engineers or been Founders that the um 57:33 there's so many that have failed in 57:35 their Endeavors because they didn't have 57:37 someone that know how to solve the 57:39 technical parts 57:41 and now that we're close to having 57:45 people know that this was like know that 57:47 this was exist and they work they solve 57:49 everything uh it's going to be an gri an 57:53 explosion of like entrepreneurship and 57:56 better software product we're not going 57:59 to settle for all the annoying bad 58:03 technology that we that we use today and 58:07 um 58:09 everyone uh who has an idea is going to 58:12 say like okay I'm gonna build this thing 58:14 and show you that this is the best this 58:17 is the best version of the product or 58:19 what our company should be doing instead 58:21 of having long meetings or like writing 58:23 up documents so it's um going to be 58:26 empowering across a lot of different 58:29 professions and and places in the world 58:32 what's what's next for lovable what's 58:34 kind of like the next few things they 58:35 might launch as this episode comes out I 58:39 mentioned this agentic behavior and when 58:41 I say agentic what it means is that the 58:43 you give more freedom to the system to 58:46 decide what what happens next it might 58:49 want to write a test run those tests and 58:51 see like oh the tests fail let let's fix 58:52 those so so that's um one the big 58:56 unlocks for getting further faster and 59:00 on then there's some more like obvious 59:03 things that you want to 59:05 do to go all the way to easily go all 59:10 the way to making money with lovable and 59:13 that's that's like how do you set up so 59:15 that it's hosted on your specific domain 59:18 how do you collaborate seamlessly with 59:21 your team H and making that that easier 59:24 so that are just obvious things um and 59:28 something we're thinking about is to 59:31 help the founders succeed after they 59:34 built their first version and like how 59:36 do they get more users how do they get 59:37 feedback how do they get the word out if 59:39 they buil something 59:41 useful I was just gonna say that that 59:44 that's that's exactly where my mind went 59:45 is like everyone's gonna be building all 59:47 these things no one's ever gonna get any 59:49 traction with these tools because no one 59:51 knows how to find users get anyone to 59:53 basically go to market and growth is 59:55 like a whole different skill so that is 59:57 so cool that you're thinking about that 59:59 how do we run some paid ads for you how 60:01 do we think about SEO how do we think 60:02 about Word of Mouth reality referrals 60:05 that is very cool okay yeah we already 60:07 have some playbooks that we that we help 60:10 the people building with like how how do 60:11 you do those things that you can find up 60:13 on a Blog oh interestingly this makes me 60:16 want to buy some meta stock because all 60:18 these apps that everyone's building 60:20 they're gonna all be running paid ads on 60:21 Facebook and Google oh my God what a 60:23 good business those other guys got uh I 60:27 want to come back to you said that you 60:28 can work on your existing code base this 60:30 actually a big question for a lot of 60:31 people they see all these tools they're 60:33 all like amazing for prototypes and 60:35 concepting you talked about how you can 60:37 actually do this within your existing 60:39 codebase use lovable let me correct you 60:41 there you cannot use it on any con 60:45 existing codebase got it um we kind of 60:48 have a research preview of of importing 60:50 your codebase but what you can do is if 60:53 you start in lovable then you can have 60:55 engineer is editing it how in whatever 60:57 tool they want to use for editing it 60:59 okay cool that's great clarification so 61:01 I guess just for people because a lot of 61:03 like most list here are not building 61:04 something brand new they're working 61:06 within an existing product so you're 61:08 saying that that is coming you can use 61:10 lovable in the future in some form with 61:13 your existing app and product great wow 61:17 that's huge okay because that's 61:19 basically the most most people so that's 61:21 going to be a big deal okay uh final 61:24 question we have the segment on this 61:27 podcast called failure Corner okay where 61:31 most people come this podcast they show 61:32 all these stories of success and 61:34 everything's going great and here's all 61:35 the things always winning you guys this 61:37 is a good example just up and to the 61:39 right the fastest growing product ever 61:42 uh what's an example when something 61:44 totally failed in the course of your 61:46 career and and what did you learn from 61:48 that I I'm withit hard pressed to find 61:50 something that totally failed but I 61:52 think there's a bit of a product lesson 61:55 um where I was the first employee at an 61:58 AI startup here in stockolm called sonal 62:01 labs and the premise was just okay so 62:05 humans learn in different ways you if 62:07 you personalize then you get two 62:09 standard deviations more uh effective 62:13 learning so there there are a lot of 62:16 products like education software that 62:19 helps you learn um that is not 62:22 personalized and we could build we were 62:24 building an API personalized 62:27 learning uh and the I mean the AI and so 62:32 was it was pretty good but 62:35 the thing that we were doing in the end 62:39 was to say like okay here's this product 62:41 here someone has build a product or some 62:43 some way to learn where be it like 62:45 English thing du lingo and then that the 62:50 people that have that product have to 62:53 use this Advanced AI API to start to 62:56 making it 62:58 personalized and it was it's a very hard 63:01 like retrofitting like oh you have to 63:02 switch out the engine and put in this Ai 63:05 and it it's the big learning here is it 63:10 didn't work very well on for the company 63:12 I mean the company wasn't super 63:14 successful in this the big learning is 63:16 that you have to start with like how is 63:18 this product working end to end and then 63:21 add AI or think where should we add 63:24 AI so that was big learning for me that 63:27 um you you really want to see like like 63:32 how the what is the big picture of the 63:35 user what's the big picture of how this 63:37 should how do you think the user 63:38 experience should be and then add 63:40 something with AI uh to solve specific 63:43 problems and now now Sons is doing great 63:46 but it's it's not on top of that product 63:48 specifically that's I think it's a lot 63:50 of people hear this and they're like of 63:51 course but I think it's so hard to 63:54 actually remember this point when you 63:55 have cool Tack and you're like holy 63:57 everyone needs to try this they're going 63:58 to love it and then you don't realize 64:00 like no one actually cares if it's not 64:02 solving a problem for them you know 64:03 there's like a lot of novelty products 64:05 that like everyone want to use for a 64:07 little bit and then like forget and it's 64:09 not I don't actually need this often and 64:11 so I like what this makes me think about 64:13 is there's all these product lessons for 64:16 what is likely to help your product be 64:19 successful and an app like like a tool 64:21 like lovable can help you do this 64:25 because because if someone is building 64:26 something you can guide them okay what's 64:29 the problem you're solving for somebody 64:32 how many people have this problem how 64:35 much does this matter to 64:37 them maybe we should add like the Lenny 64:39 mode it activates in lavable activates 64:42 like this product product 64:45 coach questions you wait hold on are you 64:49 this why let's take a step 64:52 back everyone's G be like out of my way 64:57 yeah 64:58 exactly what's your yeah what's your 65:00 experiment plan uh that's actually I 65:01 think there's actually a big opportunity 65:02 there to say people because you know 65:04 there's like a play around with this 65:05 thing and then there's like okay but 65:06 really is this anything people actually 65:09 want can we call it Lenny mode is that a 65:11 fine with you 100% awesome let's do it 65:15 I'll license you no cost sure okay okay 65:18 we made a deal here let's do it okay uh 65:21 Anton is there anything else that you 65:23 wanted to share anything you want to 65:24 leave listeners with 65:26 uh before I let you go and go to sleep I 65:28 think again the the world is changing 65:31 quickly and it's very fun you should see 65:34 that's like have fun in all all of this 65:35 change um and the best thing you can do 65:40 for your current profession or if you 65:42 want to have a new job is to be in the 65:45 top 1% in knowing how to use the AI 65:47 tools so go out there use use lovable 65:51 use other a tools and become um make 65:54 sure to understand or try to understand 65:56 as much as possible and how to use them 65:58 productively um that that's that's 66:01 something I I tell all my friends in 66:03 jally and I I like love the audience to 66:05 know as well okay well I gotta make try 66:07 to make this even more specific for 66:08 people uh how do you know if you're in 66:10 the top 1% like what's like a heuristic 66:13 almost of like slash how do you get 66:15 there is it just use it a 100 times a 66:17 day what else what can you recommend 66:19 yeah I I think if you spend a full week 66:23 on trying to reach an outcome the best 66:26 way to learn is like I want to do this 66:27 thing and then I want to use AI to do 66:29 that thing uh and you've spent a full 66:31 week you're you're in the top one% in 66:33 the global population if you have 66:37 friends that you surround yourself with 66:39 friends who who have this Obsession or 66:41 they also care a lot about this uh then 66:43 you'll be quickly in the top 66:46 0.1% so what I'm hearing is like find a 66:48 problem that that need that can be 66:51 solved like find a problem a pain point 66:52 for yourself or someone yeah and then 66:54 end to end like fully solve that problem 66:56 spend a week getting from idea to like a 66:59 thing that was actually somebody's 67:00 actually using yeah and you're in the 67:03 top one% yeah I I think at the top yeah 67:07 the top one% by just spending a a full 67:09 week and making like asking AI if you 67:12 don't understand so make making sure 67:14 that you understand yeah like that's the 67:16 thing people forget you just ask like 67:19 like you would would you ask the chat 67:20 feature of lovable in this case or would 67:22 you go to cloud or chat GPT to ask for 67:23 advice I mean my recommendation here if 67:26 you're in product is to use lavable to 67:29 build software and and learn that AI 67:31 tool if you're and then you should use 67:34 chat mode and chat mode I have to add is 67:37 something you activate in your user 67:39 profile it's not launch like in the in 67:43 the main product so it's in in in Labs 67:45 but if you add that flag then you can 67:48 use chat mod if you're if you want to 67:51 learn some other AI tool then you should 67:54 I mean ask that tool or ask Claude B 67:58 about how how how that topic that domain 68:01 works okay amazing uh where can people 68:05 find you where they where can they find 68:06 lovable and how can listeners be useful 68:08 to you lovable posts updates and me on 68:12 lovable Dev on Twitter and we post 68:16 things on LinkedIn as well and there a 68:17 lot of lot of things coming out and 68:20 changing in how we will software so you 68:22 can follow lovable Dev and you can 68:24 follow me at Anon OA at Twitter um I'd 68:30 love more feedback on what people like 68:33 where people see this is huge change for 68:37 them and we there lot a lot of people 68:39 posting about that on Twitter but there 68:42 we have a Discord where you can share 68:43 like oh this is how I use loveable it 68:45 was super useful to me um and H 68:49 feedback. lovable dodev you can give you 68:53 can ask for uh new features there's a 68:56 lot of people asking enough boting what 68:57 features you want next so and that's 68:59 useful that's the most important thing 69:00 for us we just want to solve people's 69:03 problems amazing Anon you're doing 69:05 incredible work what a what a journey uh 69:07 I'm excited to have you back someday 69:09 when we see more chapters of this 69:11 journey to as do we all that's why 69:13 people listen to this podcast uh Anton 69:16 thank you so much for being here thank 69:18 you so much Lenny bye 69:22 everyone thank you so much for listening 69:24 if you found this valuable you can 69:26 subscribe to the show on Apple podcast 69:28 Spotify or your favorite podcast app 69:31 also please consider giving us a rating 69:33 or leaving a review as that really helps 69:35 other listeners find the podcast you can 69:37 find all past episodes or learn more 69:39 about the show at Lenny's podcast.com 69:42 see you in the next episode