Gonzalo Espinosa, Double: Flying cars, hyperloop, cybertruck and AI

another founder wisdom podcast YC Edition with Gonzalo Espinoza Graham he is co-founder at Double so he's part of the winter 2020 tree of YC interesting background this guy has so we're gonna talk about AI GPT Forbes SAS startups moonshots um yeah moonshot is pretty pretty much one of my favorite Gonzalo has worked at uh Tesla um he has worked on the hyperloop projects I think I might want to start with that or the the Cyber truck he's done like so many ambitious things and now he's building a double uh which is an AI sales company I mean it's pretty broad but we're going to discuss more about it Gonzalo welcome to the Pod can you briefly introduce your weird self uh Hey Charles yeah thanks thanks so much for having me uh I think that was a good background and like you said right now working on double um AI company part of Y combinator 2023 batch and yeah I would be excited to tell you more uh where should we start yeah hyperloop um so like the moonshots have been in your mind for quite some time and the action to you know because you could just dream up somewhat like myself I think I found that kitty real though I am gonna um I will be back Terminator style um but yeah let's start with uh the hyperloop arrow and composite structures which you are part of uh tell us about the project tell us about the why and what you accomplished there uh yeah so so for context so I went to the University of Waterloo in Canada uh for the listeners who know it um and it all started back then actually so we started with making submarines for uh submarines for for racing so uh to attend a submarine racing competition which happens in the US and the UK and that's kind of like where the idea of like oh you can build things and you can compete them and there's other people in the world who want to build these things um and came to mind and of course being a very supportive environment where you know University of Waterloo wants to wants you to do things like these and you're surrounded by people who uh who not necessarily know how to do do it but like if you if you tell them like hey let's go do it they'll they'll be right there with you hyperloop was really similar to that uh which was actually I was the Reddit hyperloop team so on Reddit so of course Elon announced you know SpaceX is going to host a hyperloop competition great right mostly for universities but on Reddit someone uh was like hey why don't we do these for us as the Reddit team um and it actually got a lot of traction we we actually built a pod uh we can put it in three years we won some awards um and it was uh first of all it was I think it was at the time like probably like the most impressive crowdsource project just in terms of talent from all over the world coming together to build something and actually like deploy it and that also set the foundation for the team to after that tackle all their challenges so they didn't actually just build the hyper I was only involved with the hyperloop uh but our loop as what this it's what this group was called when to build like you know an aircraft and um and I know they got into crypto and all these kind of things um so it was more about this idea of like foreign organizations that are you know capable to like do these really challenging and for the ambitious projects right so you've been at the same University as Elon and then you later uh find yourself working for Elon as a quote-unquote hardcore intern um what did you learn from that one two did you get to speak to Elon what did you work uh what did you learn from him and then I'll have a question about these uh futuristic looking electromagnetic wipers you helped developed yeah so I think like even it's very actually very transparent with what he preaches and he actually does exactly that right so uh he often talks about he used to talk more about it but he he used to talk a lot about thinking from first principles uh and at the time like no one else was talking about it now it's a this very accepted idea that uh really preaching that at Tesla on every single decision just being like starting from the most basic principles and working upwards was one um and then other other ideas as like you know a flat hierarchy so everyone on the team um so so yeah there's like everyone can the information can flow as rapidly as possible everyone can make decisions team can move fast and really focusing on what matters basically I think that's the essence of Tesla uh that's how a small company or what it used to be a small company got to compete with like huge companies and took over you know the EV Market the automotive Market uh yeah it's this idea of like just first of all making making smart decisions reasoning for first principles and after that uh executing extremely fast um but you know just by basically building an efficient organization so those are things that uh you know Elon preaches Tesla it's a great example of that and uh those are things that I've you know taken to my core you know and whenever you're an environment that it's the opposite you know when there's too much bureaucracy uh you can really see like how like these companies who are nowhere yeah did you get to speak to the man himself yeah yeah of course um yeah we I was working on the science studio um on the so there's a new programs engineering team which basically works on those vehicles that you see on the nvl's the semi-track the Roadster uh the Cyber track you know when Elon goes on stage and shows oh this is we have that one prototype that drives around actually drives around it's a work a working car um the new programs engineering team is the team that builds those those vehicles uh and so it's a iterative process basically you know goes from like sketching on on paper what we think it should look like all the way to actually building it and and he was very involved in that so it was almost like a weekly thing uh yeah he has he has huge Presence by the way his uh he's an aura his Aura like is it stressful to to be around him I I want to say it's stressful just uh just no [ __ ] just go straight to the point right yeah as everyone should be so who was the intern that got fired over the non-bullet-proof glasses that he threw a metal ball in no no that was an intro no comment uh they actually have they actually have a really good uh glass team and Tesla has done a lot for for glass and windshields I don't think yeah maybe like a like pretty fragile not a school but if you've ever sat inside of a model X on the front roll and you see like the huge windshield uh you know biggest windshield in a production vehicle um you know it's very special and those kind of things are are very hard to pull right so Tesla has a lot of glass Talent uh and so yeah I was about to say um the Cyber truck is very much built out of first Principle as well which is like built the most solid frame and structure and start from there even if it's kind of ugly would you agree to that yeah well um I don't know what it looks like right now right as it is starting to reach production I think that the idea at first was to build a very simple body uh you know would allow you to have very simple tooling uh and so it would like simplify your your production lines and the investment into buildings for Action lines uh from what I've seen on the pictures as the car gets closer to traction it looks like you know yes on the outside it's very boxy but on the inside you still see a lot of the features that you see on a normal car especially on in the body like body and white on the structures you see a lot of 3D stampings and so um I don't know I think it will be a mixture of both um yeah I think it's a very interesting vehicle for sure um and I think a lot of uh yeah the idea that you know your exterior panels are also structural parts it's of course very um very in tune with that thing of like think from First Response yeah all right now let's talk about flying taxis um so yeah cities are pretty much held uh and the the traffic and uh airport experience or even worse um as a nomad I've traveled quite extensively and still up to this day I mean I've hacked my way through it through various um travel hacks and so forth but still uh not a cool experience so tell us about that moonshot what you sought to accomplish and what you did accomplish and if you let that one undecide or if you're gonna revisit it someday yeah so so basically after Tesla uh I got together with with a group of Engineers and we said like hey uh for context this was back in 2018 right so um so a lot of people still didn't know about Tesla but at that time yeah for the people I test like it felt like this this already happened electric vehicles are going to take over it's done let's think about what comes next and at the time it was like okay if cars are electric what's next either boats are electric or uh airplanes are electric um and like we built these Theses around like why we should have electric aircrafts um and it wasn't as much as Electric find aircrafts just for the sake of electrifying them it was more about building what we call the flying car um so a flying car is an analogy that for an aircraft that like a car you know anyone can drive right so you don't need a pilot license you don't need to go through like certification all these things it's very complicated to to become an actual pilot today um you don't need like an airport you know cars cars can take from anywhere cars you can start driving a car from your house so and the flying cars should also just take from anywhere and the other thing is that you don't need jet fuel right because jet fill was another huge Superior to your entry to Aviation very fast cost uh not widely available the weight also the weight yeah um and then there was nice right so you know airplanes of course it's too noisy to be flying around uh so we thought a lot of these players could be uh broken by basically building a small uh electric aircraft which we we really like to call a flying car and we paired that with um at the time Boeing was also interested in this idea and they had a competition called the go fly competition which was one million dollars to build this aircraft basically uh yeah so so we started working on that um we worked that on that for like about three years we started building you know the small prototypes um and for context you know we started with nothing from scratch no money and so the idea was roughly like okay first build like a very small drone uh develop software and use that to raise some money then we raise some money build a bigger version of that uh versus money to build an even bigger version of that that can carry a person but so it took three Traditions to get to something that could actually carry a person uh and three years um but yeah sadly that also lined up with you know the pandemics and personal personal staff at some point um kind of lost interest in that um but I think it's great to see that a lot of other companies are pursuing this this same vision and some of them are you know already publicly traded it feels like it's just a matter of time you know a lot of them are pretty pretty far into like regulation processes so I think in a couple of years we're going to see these uh for sure I'm seeing that uh the prizes um like for example I know xprize and yeah certainly know the prices that uh Elon puts for that has really helped you so and me the insights I'm getting out of that is that I certainly want to put this price later on to influence young minds to learn and grow and go for moonshots because there's not a lot of incentives to do so yeah I think that's that's completely correct I think um there's a lot of good ideas and uh well depending on which side of the market you are you either have a lot of money and not enough ideas or a lot of ideas and not enough money and sometimes yes these prices can help uh Bridge those gaps so so yeah that's an interesting observation and then ninja delivery so you probably went in like the hardest Niche ever uh we've seen a lot of these companies close down uh what happened there so okay so so okay so after after building these flying cars you know um again we were in the pandemic this was 2021 we were all locked um I was in Canada at a time we're all locked in our houses couldn't go out uh and it was my birthday I remember and so I wanted a cake for my birthday and so I did what we all did back then right you were on instagram.com you look for cake you order cake they say cake is coming not today in a week right but my birthday is today so that was an issue uh they got me to start listening to ways to like you know X Twitter delivery and we basically had this idea for a robot that delivers food but it was a tele-operate robot so think of like a remote control car um but the person drive the car is driving in the US or in Canada you know very expensive markets uh but the person driving the car is in Mexico um so we were kind of like we're going to monetize this weight Charter trash uh and so we spent a summer building this robot and and driving it around uh we were still in Waterloo in Canada like these like small student town and we were driving around fulfilling orders talking to local restaurants getting them to like hey like look you should use this robot it's the future instead of using Uber Eats get the robot you you get to spend the money it's way cheaper uh but we didn't get a lot of traction restaurants restaurants are actually extremely hard to sell to and then the other thing was that building robots especially back then if you remember supply chain issues and all that stuff I was actually extremely hard building Hardware is hard of course takes time takes money um but what our risk our users valued the most was actually the speed at which we delivered um because whenever you know whether a robot delivered or food it was always very fast like 10 minutes 20 minutes um and so we we ran with that with that idea of like super fast delivery very convenient and instead of robots we just switched to electric bicycles and we launched this in a student town again just around campus it took off right away it was like ninja 10 minute delivery um you know kind of like a new idea for Canada for the Canadian Market at the time we were already seeing it in Europe had a lot of traction no one was doing it in Canada so just copy paste uh and yeah like it really took off in Waterloo um and then uh yeah so so we scrapped the robots just just delivering in e-bikes raised some money expanding to Toronto which is Canada's largest city like the Canadian New York basically uh and just try to grow as fast as possible um before you know before it stopped working um super harsh model um in terms of 10 margins yeah super thin margins uh really hard to scale um and there's a few reasons why it's hard to scale uh one of them is like you know you're basically hiring like very cheap you want very cheap labor but also you want really really motivated labor for them to like really deliver in 10 minutes you want them to care um and yeah it's just a hard equation uh and at some point you know no one no one was Finding uh any of these companies anymore uh actually a bunch of our competitors are out of just went bankrupt we managed to merge with one of her competitors so they're still at uh a 15-minute delivery service in Toronto which is it's it's a great outcome in a way for for users uh but yeah it was pretty tough um the one thing I would say uh the one takeaway was that um ten minute delivery or at least the way I see a Tina the movie was was great like the idea that you click on an app deliver like get me these and in 10 minutes in your door 10 minutes or less sometimes we're even faster uh in 10 minutes is in your door it feels like magic right so we basically created this this experience that feels like magic and I think that's why um yeah that's why we had so much traction and that's one of the insights that I I took after ninja onto the next company uh yeah because I'm a power user of rappy and they are generally delivering 25 minutes really impressive startup in itself uh but two things uh two finalize on that one um one tell us about the juicy story about like someone delivering something one of your delivery guys doing something fun because I'm pretty sure you have those and second tell us how did that one close down did you had a major debt and how did you roll out uh the shutting down of that one yeah yeah two good questions so uh so so first of all for the first question so actually like so ninja did like thousands of the thousands of deliveries the first thousand deliveries were actually just me and my co-founder we're a team of two uh we're working out of our kitchen so yeah there's definitely a lot of fun stories um uh yeah so I don't know my most my most uh precious memories of Ninja for some reason was when it rained when it rained really hard and we were like trying to like oh man because there's this fun correlation which is like people want delivery when it's least convenient to deliver right so right when it's like snowing and it's hailing when uh there's a storm that's when people want deliveries so because they like to see you suffer together it feels that way so so yeah so a lot of times uh you know we made some really really good memories that way um yeah we also got a lot of tricks Small Tricks here and there to a little bit faster so for example if you're ever delivering to a to a tower a building with a lot of floors say you're going to the 10th floor you get off the elevator before you get off the elevator you send the elevator to the next floor so by the time you're done dropping the back you come to the elevator it's already way down little tricks like that saved a ton of time um and and yeah so um but but again those are will not scale as you start hiring a lot of people that frankly just want a job and they're not as aligned with the the 10 minute delivery mission right uh yeah yeah I have two funny ones to tell so the other guy I don't know why but I decided to order like just one baguette uh to get here for uh Trenta pesos and the guy came so I don't know why I did that uh maybe I was too busy or something but no the most funny one is like early days rapping I was in uh ciudad de Mexico and I ordered a full order of grocery uh paid like 100 200 something like that and the guy came with three huge ass salaries and that was it you know yeah like I ordered the whole thing and then he he pulls out of this bag like three huge salaries and then that's it the early days of Rapides were like super chaotic um I would say like probably one order out of five made it you know all the others that were canceled um yeah it was an early user of them and because I wanted that product which is another discussion but tell us how you closed that one down um how how did the ramping off go I think like what rappy did cancel when you cancel an orders that's usually a good sign right that's when the wheels are coming off and uh yeah um but uh in terms of the end of ninja basically um it wasn't as much of like we ran out of money or we had too much debt or anything like that it was more of like um we've been trying to fix the economics you know we've been trying to lower the cost of delivery we've been trying to lower the cost of acquiring customers we've been trying to increase the average order value of of like what customers are ordering you know we try a bunch of things uh and it wasn't working so we couldn't see a way out um and we were both at the hospital with hypothermia and we were we didn't have staff anymore yeah well back then yeah back back then you know we also have a lot of Staff right so you do definitely feel more responsible to to towards driving towards a good outcome right but yeah it was more of like if if there's no path forward on this and we cannot make this model work instead of burning more money we should just you know return the money to investors uh if we don't have any any other better ideas um and that's exactly what we did right acquired um is what's written there so I guess yeah and that's how it yeah yeah yeah basically we merchanted with this company called buggy um yeah interesting um how much more time do you have in your hands uh I have like 10 half an hour okay even better because yeah you're interesting um because your next step is like kind of licking your wounds uh at ecobee for a year and then coming back in the ring Like a Warrior after that with double like do you do you understand like how intense and driven you are do you you often reflect on that so so I like how you said licking your wounds uh like so in between you know in between these like moments in between Career changes and stuff like that or when your company fails you know you should take some time off to think of like okay you know maybe some maybe look inwards and things like oh what what do you really care about uh what's really interesting what's happening in the world um and at some point um I was really into uh thinking of like okay we should do something on the environment and I will stuck to all these ESG companies uh camera credit companies and then I discovered that or like so someone mentioned that um around 50 of the energy used at home it's from at least in in North America yeah right so this this would be passively different in other places of the world but at least 50 of the energy in North Market you've seen a house it comes from your HVAC so Heating and Cooling and it could be worse in a good position to like you know directly uh like fix inefficiencies with heating cooling uh and that's kind of like how I rationalized it uh I think looking looking back maybe not the best fit uh just doesn't look like a moonshot let's call it that way well yeah um right just going from yeah yeah um I actually haven't thought too much about that but uh yeah but I I it compensates for your past moonshot don't worry you're you're very much diverse in the moonshot area of things yeah I mean there's also like immigration related stuff you know like uh I was living in Canada I wasn't Canadian kind of like yeah like life happens around these decisions and you sometimes you still need to to and it shouldn't right like as a government I would make it as little friction as possible which is a topic for another time I'm interviewing duck deck Founders uh nowadays uh so I'm gonna send you a couple with that but that leads us to double which is yeah you're a software company so there's no like physical product why why did you decided to switch from flying cars to building a sales AI software yeah so two things here right so it all started with Ninja delivery actually so like like we talked about Ninja I was delivering groceries in 10 minutes is you basically have all these like micro warehouses around the city so you're very close to your to where you're delivering always uh but when you're managing inventory throughout all these little stores that actually we kind of said data problem uh you know you need to categorize your items you need to place new orders and sometimes your items names do not match what your vendors call it there's small differences like that your vendors websites are all different some some you don't even have a website they want like a form like on a PDF there's all these things that need to happen that are very manual and tedious and and happen on repetitive basis um and so when we saw gpt3 launch first uh we felt like hey this is look actually just step back our solution to all these data issues was hiring Bas from the Philippines so even at like eight four dollars an hour um which is what you know normally that's the rate even at that rate we were spending hundreds of dollars uh a week just dealing with these issue and have it constantly break yeah exactly it wasn't perfect you still need to manage it um and and you know ninja you don't normally want to manage too many uh let's call it low leverage uh employees it just takes a lot of start building a huge organization of just middle managers yeah and so anyways we saw GT3 launch and we thought okay this is perfect because it can context rise information right so uh you can spell toilet paper in 10 different ways it will know it's toilet paper uh doesn't matter if it's totally paper bulky or extra or anything like it's like it can do things like that and it can also sort of make decisions um you know it can select links based on the link name which one's more more likely to be the order link and checkout link and stuff like that uh so we thought okay this is this is exactly what we needed to automate our processes back then uh yeah and then the other thing the takeaway was the the other takeaway for me was like the user experience of ninja how 10 minute deliveries magic a lot of these AI tools feel like accepted like that like magic uh so now we had a problem uh and we had an experience when I delivered um and so basically that's where double came came in and that will started as a very simple tool which was like a spreadsheet with GPT uh three embedded in it right so you could run you could put data and you could run it through gpt3 in bulk right so you can have thousands of roles it will clean it up and have like I don't know names still funny it will clean it up and things like that and that that was like the very basic version of double and we just put it out there and people started using it uh and since it's been it's been evolving um so now if you go and use double.com it's more of like a elite generation and prospecting tool just because that was like the main like our first early Outdoors for sales team since we started adding new tools we started adding web search uh web scraping we started using a poll Integrations API Integrations uh so that's that's what double is going to now interesting and why you know why do you think this can be a moonshot because me and my sales companies and yes I'm uh thinking of getting well not thinking I'm acting upon my AI SAS and building something as well obviously not at the speed that you're doing in the full time and the YC stuff but my thesis that this will be a major Cash Cow if I can have agents making calls for me and perfectly written emails all the time saying why Gonzalo and Charles should connect for example that is piles and pies and piles of money which I can use to finance my moonshots yeah um so what's your your thesis on that first of all I really like this vision of like agents doing doing the selling at some point I will probably have agents telling to agents and uh yeah um I think what we're actually building you know you know right now we say we're a lead generation tool that's not actually what we're building what we're building is what we believe to be the IDE for like software 2.0 so these future were like basically anyone can develop you know software uh just using plain language um or anyone can integrate apis you know uh and things like that so in a future where everyone's building software what is the interface people are are using this software to build these automations and we think that's going to be double um no no like our next step here is what we're actually doing is like we're building we're making it extremely easy to build uh AI driven automations and workflows and so how does that become huge well I mean you know we can resonate from from first principles um but we can also look at tools like sapier for example so sap your you know huge company um and they basically built a mode around having all these Integrations right if you go on sapier I want to say they have around 5000 Integrations right now with different tools you know it connects to slack to to your Gmail to whatever you want and that was like for a long time that that was the mode but today where we're standing right now that mode is eroding right so uh I think we've all seen the demo where you you gave gpd4 all the API documents for for a certain API let's say the slack API and tell them like oh I want to build an app that integrates these two programs and it will do it um so so basically that mode of like oh my my tool integrates with everything you know well kind of like now every tool will be able to easily integrate everything uh we believe we're we're more value can be provided is the environment that makes it easy to to you know to create those custom solutions for your needs and that's what we're building my two cents on zapier first their open AI uh Integrations plug-in is weak second I've been a power user of zpi three or four years ago and I've had some what good experiences back then but nowadays my brain what it's telling me is that zapier is equal to headache it always breaks it's [ __ ] slow okay that's another thing everything is slow and then you need to build these multi-steps which brings me to the next point which I also know that you have in your brain I I just know it um the ideal product um for that would be me asking GPT to pull me the list of all the guests of Joe Rogan and give me their emails and boom it replies with the goddamn CSV end of story I'd pay at least 100 through 150 to 500 bucks per month for that solution what do you think about that yeah no I think I think that's that would be great right and I I don't think we're very far away from that I think we're actually we're just around the corner like because instead of building another clay why don't you do that um with double yeah so um so I think that like without saying like oh that was gonna build agents um because we don't call it agents but that's basically I think the next step so what we're seeing on when people come on double um you know obviously there's a limitations of what you can do with you know the gpt4 API and you know these tools would build web scraping Google search email stuff like that but people come with higher expectations and so we can see uh we often see double fail meet the expectations and sometimes you know people are gonna say things like oh great at least of everyone who has been on Joe Rogan uh and I'll put their email kind of thing right and we can tell like oh we're failing at it uh so so right now the next step for us is to to be able to like yeah just go from something as like that and be able to automatically create a workflow and also know when we're failing um nothing that you've said is unrealistic right like um a person could do it uh so so we we think yeah we can do these kind of things yeah I've started seeing hints of that with the PDF uh plugin so you scrape them with uh Big Data scraper whatever Chrome extension it's called let's say on all this Google Chrome podcast yeah yeah and then you download CSV convert it into the PDF and then ask uh judgpt to come up with all the names and then you ask the the names to be associated with the company you pull them out in Apollo and yeah so we're starting to see that so yeah so you've laid out a few processes you've also mentioned tools that can read about like PDFs and interact with the PDFs uh I think you know the fact that you know chat with PDF is an idea that you know you could just mentioned it but there's already so many Tools around the acidity of oh chat with PDF Challenger dolphins and stuff like this uh so I don't think those are the kind of tools we might developed because at the end of the day it actually matters to build a mode uh around whatever it is that you're developing uh you cannot just simply build these these rappers right so what would be your mode there uh so what we what we're building towards is um so double is again this tool to like build automations build workflows using Ai and we want to you know uh once you set up the workflow just keep using the workflow right it already works it's hosted on devil uh why would you change it anything else uh yeah just you become basically reliable on this workflow that works great and then you can build more workflows on top everything's in one platform you keep coming back to that uh yeah so it's a very old-fashioned kind of mode um it is the definition of mode oddly enough because um people speak about data modes and all that [ __ ] but let's not forget that mode is brand brand is human psychology human psychology is myelin sheets around neuron cells and their strength you know how strong is that connection when I think of zapier I think headaches you know when I think chat GPT I I think pure serotonin um so that's a mode right there I know that you're looking at the time here I just want to wrap up um well well first of all I think it's yeah it's funny you say stop you're eating headaches I think it's happier I think like oh money burning it's so expensive uh at least like at previous companies that we've used it we became so dependent in these workflows and they were so expensive but again we were so dependent on them that we'll just pay whatever it will be um right and it's a similar idea with uh devil basically you know people set up these workflows these workloads are very capable I don't think there are any tools out there right now that can create the same workflows we can create but I do think that you know it's just a matter of time to like you know other tools that can create these workflows appear but by then you know like we've already have these workflows created for you and we'll provide more how to improve them so so yeah that's that's the idea for double yeah the price headache is Clay your competitor by the way it's a quick hand for you um I've talked with Eric yesterday um he is gonna deploy also a copy based type of template which is something needed right because Eric shows all of these great um Outreach and what's possible but I think what we need is just a copy-based template yeah I think template um so so again going back to this idea of like oh 10 minute delivery was magic uh you've seen an AI tool should feel like magic and if if you're packing AI but you're not delivering a magical experience you're you're kind of using it wrong because at the end of the day like the most basic essence of you know these AI tools is I would say it's tragic Theory right that's the entry level and that one feels like magic and if you look at all the other companies in AI that are doing really good like copy AI Jasper uh and so on like basically you on board as a new user and you directly get the value like you get a great experience there's no learning curve you you get it right away um and and yeah so so that's that's what double should be too and and yeah you think like most products are gonna be chat with like uh charging PT's uh serotonin hits yeah so actually um I think I have personal biases versus chat and I think I've been proven wrong it seems like chat is taking over a lot of interfaces um at first honestly I was accepted I accepted I thought chai GPT would be the chat to rule all chats and I think I've been proven wrong um I'm seeing you know I've seen a lot of chat products but a lot of them seem to have um a lot of traction and I'm also seeing when given the choice to interact with your interface or a chat uh users are flowing into just you know going straight to the chat not having to figure out your interface just saying what they want and extracting value so so yeah that's that that's interesting I think what I would launch is a community of people in the probably the cells industry that would constantly suggest new uh automation framework in the cold email industry and email out LinkedIn Outreach industry with what we're speaking and so so that like people could suggest new models and the center of that would kind of be a chat that could participate in their Community to some extent and it would gather all that data and it would build what the users would kind of want anyway um yeah if you think about that one I think it would be really cool especially like you know um because right now uh so in this community it would be really cool to have you know a chatbot that it's not only like reactive it doesn't only take commands or like you know a lot of information but it's also like proves you and like ask you questions and stuff like that uh yeah I think that will be a pretty cool uh something really cool to build and also a really important market research Tool uh at least right now even ourselves we're still doing a lot of you know people coming will post things and we're still have to like manually hey like ask them about them extract more information and then build build it based on the feedback yeah once you automate that um yeah that's gonna be very powerful because AI being creative and really advancing and R ding quote-unquote is still not there like I'm not sure if you share that feeling or that's that's the missing piece right to AGI yeah um yeah that's that's interesting um yeah I I feel like students are still in charge of that basically which is why humans will feed the r d and this butt will actually build the the technical and the realistic part for them yeah I I think you know well so the the only the only part of double where we see a lot of creativity is people using it to generate like personalized messages and yeah there's a lot of room to be creative and yes the the vanilla output of gpt4 is usually not what you're going for generating messages that are not something that a person would ever say uh or it's you know too cringe or it's like it's it's just awkward yeah you call it vanilla I call it cheesy yeah yeah it's exactly it's very cheesy right and but that's where opportunities to again like really give examples uh and get the AI to know your what you you want just write more context um like well explain exponentially improve your your output like it's really that's really how you should be using that tool and again just going back to like building the interface that people are going to use to build Automation and stuff like that that's one of the features that devil already has right now oh like um you know the ability to provide as many examples as you want in a format that you know the AI is going to understand uh yeah this is a message that I I always tell users um I think people are not used to like you know having to like um you know provide all these details every time and and maybe part of that is the fact that you know jjp doesn't have a memory Beyond one conversation and it doesn't remember oh you know this is what an email should look like you know um yeah again early days I'm sure uh this will this will get solved are you sure you're gonna build something amazing I have one last question for you um so this is a brain tester um I've yeah I've seen your your LinkedIn profile there I've clicked on ninjadelivery.ca and it says 404 page not found so us Founders we have these quote-unquote failed startups right on our LinkedIn profile what would be a business idea that you could produce with that yeah uh well actually we already had a business idea right so we're um we're selling that traffic because we still had a lot of people looking us up on Instagram on social media on our website uh we could just redirect that to to literally our competitor um you know a lot of traffic so why not build a platform aggregating like all of these um websites from Founders that they're they are on LinkedIn mostly and have some sponsors in there some [ __ ] yeah no that's really interesting actually you know what someone actually beat us to uh because for web fly one of my first startups we used to have wi-fi.ca and someone already did that and it's basically uh I think it's an aggregator for you know like flying car flying taxi uh articles great idea uh and also pretty easy right because when you start are going are closing like they don't really care too much about how much they make on that domain yeah I guess this is like more uh Go Go Daddy's grounds or if they have an accelerator or something or something there might be things all that that was a fascinating interview really interesting and yeah I believe in you I think you're going to be quite successful with that one as well uh where can people find out more about you and double yeah well you just go to usedouble.com uh we we have this mission of always answering to every single message in five minutes so if you message us on like our intercom bubble on the bottom right we'll answer in five minutes you can test that I'll bug you out yeah uh and then I'm on uh on Twitter I'm just GPT on Twitter that's my that's my handle dude dude let's start another startup for Twitter handle and sell them yeah Elon grounds but okay interesting I'll follow you on Twitter uh so Gonzalo Graham on LinkedIn Gonzalo Espinoza Graham on LinkedIn use double.com uh that's a wrap my man thank you