Robin Allenson is a serial entrepreneur who studied AI at University of Edinburgh in the early 90s, and went on to found a series of AI-powered ventures. His current venture Similar.ai uses AI to automate the important and often laborious process of optimizing web sites for SEO.
Back in 1990, Robin Allenson was studying AI at university and he fell hard for the vast potential of this technology. So Robin went on to build several AI startups, including similar AI's current project, an intelligent SEO platform for corporate websites. Robin's been in the AI game for over two decades and this past year he'sseen a major shift:
companies that are actively seeking AI solutions rather than dismissing them is too risky. There was a joke in AI , in the nineties that all of the working AI models were called software. because you do some research, you get something to work, and then people would just assume it was software. Cause software was actually the thing that worked but now we have inbound from people who are searching for ai for seo, ai for internal linking, ai, for new page creation and topic research. so it feels like the whole world has changed around us. Join us to get an insider perspective on the latest AI developments, and discover how Robin found success by defocusing on tech and understanding his customer's pressing problems and struggling moments. Thank you Robin for being here. I met Robin years ago when we worked together, on an earlier version of his startup, which he's gonna catch us up on. Robin is one of the smartest and most inspiring and, forward thinking people I know. He's been working in AI for a really long time. And I think he's gonna help us understand what's going on right now. That's very kind of you, Amy, Joe. I don't know, if all of that is entirely true, but, I'll try and live up to that., it's like the blind men in the elephant, right? Like we all can only speak to our own, experience. So, start by please, Robin, introduce yourself and tell us about the startup you're working on now, who it serves, what it does. And then we're gonna drill into some of your background. Cool. Thanks. so Hello, I'm Robin. so I'm the CEO and the co-founder at Similar AI and Similar ai, we're automating SEO for agencies and for in-house teams. So search engine optimization is a very manual job typically, and there's a lot of folk who are, working perhaps in a large enterprise site or perhaps in an agency. And, mostly they're taking some of the existing SEO tools out there using their user interface and spitting out some data which they throw into an Excel sheet, or a Google sheet, and spend a lot of time manually updating that and trying to match that up to their pages to take some kind of action. and what we're enabling is basically teams to, to use our no-code, tool, and toolbox to automate that, to speed it up. So , we are kind of, taking a those manual things and sticking all the data together, and we call that data ingredients. And on top of that, we allow our, users to run, recipes we call them, which are no-code programs, which can publish and update , their websites to make them more user-centric and grow organic revenue. so that's, kind of in a nutshell. I could talk for days about it, but I'll try and keep it succinct. So who's your core target? Who needs this the most? so, for the longest time we've been, focusing on, in-house SEO. And typically there's one SEO working for one, search engine optimization, product manager working in a big enterprise. the site has perhaps hundreds of thousands or millions or tens of millions of pages, and they, on their own sum have to update the site to match it up to how people are actually searching and make it more intuitive , for search changing users or as we call them nowadays, people, and try to make the whole site work. and so that's just a monumental task. SEO is normally something you do by hand, and so it's just incredibly hard to do that. Normally the way you solve that is maybe you get an agency involved. And you get a lot more hands, on deck. or you use some in-house, tools and you're using, you're building that yourself. But when you use in-house tools, mostly you have in-house data. And so that becomes a struggle to actually match up to how users search on the internet. so that, that struggle means that a lot of those sites end up with too many pages, and those pages aren't interesting for search engine users. so that's the struggle that we're trying to help and we're trying to turn those users, effectively. So one of our customers, they said that we give our, we give our users SEO superpowers, which is, I think it's a great way of thinking about it, using ai. Cool. So we're gonna get to how you came up with this, more about how it works. and before I get into your background to figure out, you know, like sort of how you got here, I can't resist asking about the relationship between SEO and Chat GPT and the explosion of chatbots, because a lot of people are saying, wow, who needs SEO? Websites are dead. Why are websites even gonna matter in a few years if we're all using something like Chat GPT instead of a search engine? I'm sure you've heard the same argument, right? What's gonna happen to websites with all this stuff? Is it, you know, what's your perspective on this? You're right smack in the middle of this. So I think there's a few things, which are kind of, interrelated, and where this is impacted. So people have been saying SEO is dead for many years, and it, it hasn't died yet. and, but there are a few impacts where, that the ChatGPT or where large language models, could, fundamentally change, what the SEO industry is doing. one of those is search results. So Google has this Ste so this experimental approach to integrating, large language models in search. and so, the initial feedback on that is it, is that those results haven't been superb. large language models have a number of, weaknesses, such as their ability to, yeah, I call it confabulate. so, confabulate is a, a term kind of, meaning they bring up false memories, is what happens in psychology commonly called hallucination. So these things have a tendency to lie, egregiously, and in search results, I think that's a very different kind of intent than when you are, when you're chatting to ChatGPT. that can be a very bad problem, for certain types of results. I also think that right now, ChatGPT is a toy that's being used in a lot of interesting cases. And there's kind of an adage in product management is that a lot of the best products, start from toys, right? and so, maybe it will grow into that if we can solve a lot of these, hallucination problems. but then, So that's the first piece is it's already a disruption in that we see that Google is changing the front page of search in order include AI results. and actually messing with not just seo, but also, the revenue, that it gets from an ads, right? So it's willing to sacrifice that, to contend with this potential, competitor in the form of OpenAI and Bing. That's now, I think there are still a lot of unknowns, like will that actually work out? our hallucinations a solvable problem, for web search? Don't know. we don't have a clear answer on that. So the assumption is kind of, yes, it's all gonna work out, but it doesn't, there are very large categories where they just can't use this kind of approach. You already see in a lot of, like your money or your life categories. So those are things where you might give medical or financial, information. they're not using AI there, cause you can imagine that the risks are too great. so it could be that actually the number of places where you use AI shrinks over time, where they're more focused on which kind of answers they can, reasonably and rationally give. so the second piece of seo, disruption is it's a lot easier to, to produce content. and so the internet kind of disrupted the ability to publish. content that's been generated, large language models and, AI is now generative. AI is now disrupting the ability to come up with the content in the first place. and, and so that's much in the way that the internet democratized access to publishing information. so, large language models and AI democratizing access. So it's no longer just say, copywriters who can write copy, but everybody can write, copy. and so that, means that there's just an awful lot more competition. and so that's gonna make finding great answers harder. So a aka web search. so that's another form of disruption. And then the third piece is there are a lot of, folks in the SEO industry who make most of their money from generating content today. and so when the price of that effectively drops to zero, it's gonna be much harder to make money doing that. It's always been hard in a way, because, what you were really paying for was amazing, content, excellent content, fabulous content you can get anywhere else. And so I still think that there will be folks selling that kind of content, and it'll be hard to write that with ai. but, yeah, it's got harder to do, because there are lots of incredibly cheap, almost free alternatives out there using ai. So those are the three kind of disruptions I think. Wow, that's really interesting. That really helps me understand the landscape. Thank you. So let's wind it back and can you explain to us how you got from tinkering around with software as a kid to studying and developing AI and becoming a serial entrepreneur? yeah, when I was, I wanna say seven or eight, but, I'm probably getting the timeliness all wrong. But my dad came home, one on Christmas time with a computer called a ZX81. So one of the earliest computers. And, I started, as a kid to learn, to program. so I'm thinking ZX81, probably in 81, so I was probably 10. but then, I started to learn to have a program in basic, and then in Assembler, when I was about 11, I read this book called Gödel, Escher, Bach by Douglas Hofstadter. and I was smitten. I read that a couple of times. I just completely fell in love with it. and I started writing programs as a teenager to do things like, translate. Between two languages. So I'd learned Esperanto as a kid, and so I tried to build a machine translation program failed miserably, but it was a very interesting, experience. and I also built a program as a kid to, predict which horse was gonna win the race. So you could send away in those days for books, in the post I sent away for a book, explaining a horse racing system where you gave different points depending on how the horses are traced. but then I took that system and I turned it into a computer program and I would enter in the, the horses that won the races, and which horses were coming up each week. so you could just, find that in the newspaper. And I tagged that all in. And, so, that program worked and it was profitable. there were two downsides at the time was that, we made about, we would've made about a 15% year betting tax in the UK at the time was 12%. So that wasn't, amazing. And then, yeah, I was a minor and, my parents didn't really want to go to the, to the bookies every day. so that project kind of failed, but I took those, ideas and, aspirations. And then I studied, ai, artificial intelligence and computer science and, uh, joint honors at University of Edinburgh, which at the time was the only place in the UK where we could study ai. now, AI is, everywhere. and then, at, yeah, at the time my dad was a research physicist, and he kind of said, I want to do AI and linguistics, and he said, don't do AI and linguistics. You're never gonna find a job. there's no real future in that. So I took his advice and did a joint honors with computer science. but like a couple of years after that, Google was hoovering up every AI and linguistics PhD , in the country. but then I didn't, study a PhD. I went straight into, programming's job and gradually did some different jobs in, programming, kind of technical architecture, and then, kind of running, larger commercial teams, and more and more on the kind of, online business side of things. And then fast forward to 2009, did my first startup. and then, we worked on lots of really interesting machine learning problems. that didn't really pay the rent. and then we found a really boring software problem, that everybody wanted, and we started to scale very quickly. We got bought out by a competitor of ours. so I took some of that, money and then in, 2016, a long, long, long time ago, I sometimes say when at university I studied AI when it was mostly philosophy. it's funny to think that a lot of the essentials of what we're doing now with large language models, the way that, AI works way, deep learning works was still what we studied at university in the 1990s. So back propagation and neural networks, still the principles are the same, but we've had a few, fundamental restorations in how the algorithms work. Like reinforcement learning as an example. but also we've got, I don't know, a trillion times more compute and data. and so that's mostly what's actually, I think given the power into, into current life language models. anyway, that, that was kind of a whirlwind back and forth and kind of all kind of other things that, I dunno if I actually answered. It's fun though. It's interesting that there's a whole thread of probability running through your story. Yeah. Messaging probability. there's some inevitability to, to ending up doing what I'm doing. And, interestingly, my, 16 year old is about to study statistics and she's like, what's it all about? And I'm like, well, probabilities. Yeah. It's, that's kind of it so much probabilities. Right. And you know, when you talk about reinforcement learning, it's, you know, sophisticated messing around with probabilities if you learn to think in, in kind of in more dimensions, a lot of this is kind of, if you can think about weight spaces, A lot of the stuff we were thinking about then was how you would think about an AI system now. so I was gonna say, so my son is studying AI at university, today. So my son is, 20. and I was kind of surprised that a lot of what he was studying, I recognized what was going on in the textbooks. and it wasn't something fundamentally different than the AI that I was studying. what is it like, 20, 30 something years ago? and so the fundamentals have really stayed very similar. It's just do an awful lot more with the compute and the data we have now. and there are some, yeah, there have been some big leaps forward, but those big leaps forward are still, still based on, , on the same ideas. It's pretty fascinating. So we'll see where this whole craze goes, right? Like in a year what's gonna be happening with. Ai. You nailed it. There's so many unknowns, I mean, like you also said it, there's a continuity here as well. It's not like large language models have popped out of nowhere and a suddenly this amazing thing. We've had, oh no, we've had kind of big data being a thing. We've had AI being a thing. We've had deep learning being a thing. We've had reinforcement learning being a thing. We've had gans, all kinds of other kind of things where people were going, wow, this is the moment. Now we have large language models. I don't know if, I don't think the large language models are the moment. But it is pretty clear that there is a, there's a line going through all of that, where things are getting more and more capable. and the downsides to, to these kind of connectionist systems, still remain, but many of them are getting improved. yeah, so I dunno if this is the moment when AI kind of, let's say takes off, but then, it does appear to be taking off in a way that it hasn't in the last, six months. but I expect there'll be another evolution of AI in, in a few years. and it will just keep on going. So as someone who's been inside of this and wrestled with the difference between the promise and the reality of ai, what is your take on. The doomerism and the, oh my God, I'm gonna quit my job and go around the country telling everybody how dangerous this moment is. there's obvious things around, misinformation. Right. what is your take? How do you feel both as a technologist and just kind of as a human and a parent? So, so I think, so there are a few different, parts of my answer. right. So one of, there are folks in the industry who've been studying the risks of AI and how to make AI fairer, and work better for, everyone. and mostly what I see, they've been kind of, I dunno if it's literally fired or just ousted from, the big tech companies that we're employing them. not a great sign really. and then, but, and they are really looking at say, how. generative AI is using deep fakes for misinformation, uh, campaigns in actual stuff that's going on now that can incite violence. That's scary worrying, stuff that I think we could be, actively doing something about now. But I see people not doing stuff about that and talking about how, AI is gonna bring about an extinction event. So I think that seems a bit strange to me Also, , I would prefer to focus on the present day dangers. and how, LMS can make yeah, for instance, misinformation, but also bias. So these models concentrate the bias that's naturally occurring in data, which is data that's trolled from the internet. and, that concentration, I think, can be very risky. and we see models hallucinating in ways that could be, could be bad for people. that doesn't mean that, that we're gonna have an AI that's gonna be Skynet and take over the planet and kill everybody. it's just more like, this is a very powerful tool, that can be misused and it is currently being misused. so I, I think those things are risky. A lot of the guardrails that have been placed onis, and large language models, yeah, like there are pretty easy ways of getting around them. and, it's not clear how that, could work or should work. I think that, I think that, say OpenAI calling for some kind of centralized agency, to look at AI risks, is also a little bit self-serving, cause I think that's gonna push towards, more centralization. only the big tech companies will actually be able to afford to, conform, to whatever regulation is put in place. It's incredibly early in large language models to be thinking about that kind of regulation. and I think there's a burgeoning open, software scene for large language models, which will be effectively killed off by that kind of regulation, which seems pretty handy for big tech. Right. who are mostly interested in selling those, models. That's also maybe the reason why the folks who are researching this kind of, the potential misuses and what could go wrong, when they actually wanted to publish the results of those. Yeah. Big, big tech broadly. So I think in that case, Google was not very interested in them doing that. and so that was the conflict. so I think it just feels very, it feels a little bit one-sided right now. and so I'd prefer to listen to the people who are actually, have been studying this for many years rather than, so there are a lot of kind of, one of my first, my thesis at university was looking at some of the work that Jeff Hinton does. he's an amazing researcher. I don't know, just because he's an amazing researcher in AI means we should listen to, his ideas, on like what AI will bring. I don't know if being a great science researcher means that you are good at thinking about the conflict, with the science and AI and technology and, and society. there are other people who have been studying those things. I think we be giving more, credence to what they say. yeah. So again, long rambling answer. I dunno if that helps you. Well, I like it because it helps me think about the space, which is what we're all trying to do is have a position, right? Because so much info coming at you all day, every day. It's overwhelming. So many certainties, right? I found it all. Like, I don't know. And so I think, from what I read, there was a questionnaire going around asking Ai lumin and, researchers. The probability of some kind of doom scenario. And the vast majority, I don't remember the figure, 90 something percent just abstain from answering that cause it seems unanswerable. and the, you know, 1, 2, 3 something percent that did answer it, then they summarize those answers and then suddenly, that's what people are talking about. I think it'll be more useful to talk about like, that these are unknowns, rather than these are quantifiable in some way cause they're not. and I think , forcing people to quantify it seems, seems strange cause then obviously you take some tiny percentage and multiply it by infinity. and I think there was some kind of, pascals paradox, something like that, right? so you end up with, oh, we have to dedicate a lot of resources to that. I think. We should be dedicating resources and most of the people who are, campaigning for that are the folks who are actually founding these, these large language model, companies in the first base. they're also people who are using all of the crawl data without thinking about, say, copyright issues, or thinking about, you know, how they're, displacing some of the people, who first came up with the content. and just publishing that as quickly as possible, publishing those models as quickly as possible. and then saying, Hey, we should have some kind of regulation. it, yeah, it feels very one-sided to me, like, as if we already know what the problem is and how to solve it. I think we should, look at the people who are, like you say, talking about the space, and talking about the space of, of questions. yeah, I think their questions are really interesting. I don't purport to have any of the answers, but I think the questions are a lot more interesting. Yeah, I think it's something we're all trying to understand. We had, Douglas Hofstadter, the author of Gödel, Escher, Bach, which you mentioned earlier, as a guest a few weeks back, and he's pretty upset and depressed about it all. Yeah. I couldn't get a positive angle out of him even though I tried. But you know, everyone's got their own point of view. And again, just cause you wrote a bestselling book and worked as a researcher and as a theorist, doesn't necessarily mean that you have insight into everything. But I like that you're in there working positive. The reality is the book's not written. The final book on AI is not written. We're writing it. We're all writing it. Indeed. So, I'm glad to be able to hear your point of view and, you know, give it some oxygen. So I mean, take it with a pinch of salt, everything. Like my curse as a founder is to be perpetually optimistic. and so I always think the world is slanted towards, good things and a happy, hopeful future. so in my head there's always those kind of good things going on. So I tend not to focus on the, cataclysmic, potentials. I also just think we have plenty of those. Like, I heard a quote, uh, this, um, Zi Maza wrote a bit about this, and, he quoted in his newsletter, somebody else, saying, climate change should have, um, AI's, PR team. I think right now being so sure that AI is gonna bring this terrible future when we have science talking about climate change, like with this. Enormous impact. I feel like there should be some enormous evidence, that, that this is really coming for climate change. I think we have a lot of that enormous evidence and we should maybe be focusing on that rather than this very speculative, stuff. yeah, I would prefer to focus on the stuff we, we really know about and we think we can actually change, rather than this, it feels way too theoretical for me. Spoken like a true entrepreneur who likes to get his hands dirty. So, I wanna follow up on your entrepreneurial mindset that we're talking about. You're a serial entrepreneur and having now worked with a number of serial entrepreneurs, there's something qualitatively different doing it multiple times versus the first time. So from your own perspective, What are some mistakes maybe that you made that you, perhaps you see other entrepreneurs making common mistakes, especially for first time entrepreneurs, for folks that are interested in maybe having an awareness of those, what are some of the things to watch out for? Both for AI entrepreneurs, but also entrepreneurs in general? I was gonna say, daughters entering living rooms when you are, trying to, have a conversation is one that keeps on coming up. But, um, so, I think one of the things that looks like, initially it's gonna be unique to AI entrepreneurs, but is actually, very common across everybody. And it's also for me is, falling in love with your secret source. so that's kind of focusing on the solution space rather than the, problem space. And I think, there've been a lot of, recent AI companies which have exploded. Which are something like, a thin layer on top of t. And, kind of the answer has been AI is gonna solve that. I remember when I first started, similar ai, we were starting to use, deep learning and we had angle on how to create a lot of training data, and come up with, more training data means basically better models at scale. and so we had an angle on that, but we had a couple of, advisors to the company and they said, but Robin, deep learning, you don't need training data cause it's deep learning. and so they said it in, in a way that made me think that deep learning was actually in italics when they said it. so I hear something very similar about large language models now where people are like, what does your company do? We do large language models. yeah, but what, I mean, what exactly do you do? And so I think some of those companies are being disrupted effectively by ChatGPT now. and also. a lot of folks are finding it very easy to integrate, large language models into the existing incumbents. and that's actually, it's kind of harder to just do a, I dunno, an AI startup. so we fell into that. Initially, we were building a lot of trading data. We'll come out with a whole bunch of, ideas about how to, come up with large, well, basically multimodal, models is what we're building. So combining, images and texts to understand product pages and turn them into the language that people would use to search for them. and so that was the core thing we were doing, but we actually found it really hard to sell that, because we found that a lot of other people were claiming they used ai. and so we over time just stopped talking about ai. We didn't bring up in the conversation, it was in our names. It was a bit, kind of hard to avoid, but we just didn't bring it up in the conversation. And we started really focusing on the problems that our customers were struggling with, but our users were struggling with and kind of explained how it was different, and how we, help them make progress, in, in that struggle. And that was a night and day shift. and so what's been, miraculous to me in the last six months or the last nine months is actually, there has been a shift , in how people are approaching us, because now, like. I know three years ago, five years ago, if we actually, and certainly one of the first a studying ai, people would say things like, why do you do that? Because it doesn't work. And so there was a joke in AI in the nineties that all of the working AI models were called software. because you do some research, you get something to work, and then people would just assume it was software. Cause software was actually the thing that worked a few years ago. If we talked about using ai, which was in the solution space, then folks would just be, yeah, we're not looking for anything innovative. We're just trying to solve this problem. Right? And so for them, AI was something that the, no, the business innovation manager did. and not something that the e-commerce team, used. But now we have inbound from people who are searching for ai, for seo, ai, for internal linking AI for, de-duplication, ai, for for new page creation and topic research. Those are the kind of things we do. We also do AI for content. but you know, who doesn't? so, it feels like the whole world has changed around us. so now people are coming out to us and saying, Hey, we found out that actually, you guys do AI for seo. Well, you guys do AI for internal linking. We're like, yeah, we do. That's amazing. Can you explain it to us? Whilst nine months ago, if we had said that they would've ba basically gone, why do you wanna use AI what's the point of that? Now, their expectations about what AI does have shifted, and with those expectations, yeah, it suddenly means we have a new and much larger market. Again, I feel I answered two of your questions in that is, I wanna reflect back a little bit what you said and really highlight it. Cause it's such good advice. One, don't fall in love with your secret sauce. That's so hard. That's really hard. But what great advice, because customers don't care. They don't care. They just want you to solve their problem. Exactly. Yeah. So in my first startup, I had a startup before this one. We worked on lots of interesting machine learning problems. We were so interested in, in, those machine learning problems. They weren't customer problems. there were machine learning problems that we could apply and we thought there were great solutions. And then, when we found, and so, those are building kind of sales reporting. and then, one of the reports talked about a problem we found we could build some software to solve that problem. And it, it wasn't easy, but it was pretty easy to get off the ground. Once we got that going, suddenly we found, a lot of those customers were super interested in that problem, in solving it more simply. we were off to the moon, right? in the second startup. So in similar ai, I basically fell into the same trap despite advising other startups. So despite saying you really should be asking, you know, what problems your customers are struggling to solve, and have you thought about these kind of questions, and, kind of, teaching them about how to think about that and then in our own business, I've found it very hard to take that advice myself, basically cause I don't think this is something you can simply learn. I think this is something that is often, you can think of it more like a cognitive bias as a founder. Even if you know intellectually what you're supposed to do when you get up in the morning, you're in love with the solution again. and so you need to actually correct course correct each time, to go back. but when you do, it's enormously valuable. So, I've got one more question actually, I've got a bunch more, but I'm gonna only ask one more So, um, you're talking about understanding what problems your customers have and that you struggle with that as a founder, even though intellectually you know how important it is, by the way. Me too. I struggle with it too. So, what are some of your go-to methods? Robin, you mentioned that you like the magic wand question from the game thinking methodology. Can you just share a little bit about how you've used elements of game thinking or maybe that particular one to unlock customer value for yourself? Yeah, sure. so, so we actually, we did a big piece, but it was a few years ago before we're actually working on, what we're doing now, to talk to fashion designers at the time. and we were planning on building a model using gans, using generative adversarial neural networks. But they turned out to be kind of too bleeding edge. but what was amazing to me at the time was, I was able to reach out to, to, some fashion designers and we got, a lot of responses, to an initial survey within a day or two. and well, initially we didn't, and then we tried again a couple of days later and we got a lot of responses with some screener questions. and actually, some very quick five minute, kind of 10 minute interviews. We started like hitting gold, very quickly when we started asking, you know, what are the things you struggle with? and they started telling us, and from 10 conversations, I don't know, eight of the things they were struggling with were the exact same things. And we were like, what on earth is going on here? And so we kind of fed in love with those problems. and we got deeper and deeper into the conversations and we started talking more. That was an amazing experience, mostly because the value we got to, in the time we spent on it. And so, it was just a such a short time compared to building a product, which we'd already done and spent a huge amount of money on and wasted. compared to building a product, this was remarkably easy, right? And then later in the journey down, now, into, when we're doing what we're doing now, we had really focused on these big enterprise teams. So science that might have, I don't know, 150 million pages say, or hundreds of millions of pages. And, we thought, wow, this is a big market enterprise seo, talking to heads of seo, in these big companies. and we started doing some market research where we tested smaller sites. that only had I, no, a few hundred thousand pages or only had 50,000 pages. And we kind of asked 'em about the problems, like at least the way that we marketed the problems that we solved. And they were effectively blank stares. And they didn't really get what we were talking about, but we continued the conversation and we started asking about some of the problems that they struggled with, and it all opened up again. Right? And so when we actually really focused more about their, about their experience, what we found was that everybody struggles with, with automating seo. they don't always think about it in those terms. and so the way we were talking about it, they just didn't get that at all. but they would say something like, wow, internal linking is like, we dedicate time to that every month, and it's painful and it's tedious. Is time consuming and it's incredibly valuable. isn't there a way that you could like automate that and we can still get the value? and so we could talk about what the value was. They were doing it like every month they had people working on it. They, like, they knew exactly why they were doing that. and what that was true for a site that had a few hundred pages in the same way that it was true for the fact they had a few hundred million pages. and that was kind of a light bulb moment for us when we were like, oh wait, these guys have the exact same problem. They just don't, express it in quite the same way, but there's a way in which we can, kind of reframe that job, where suddenly it applies to this whole enormous market basically to everybody trying to, everybody trying to do, customer acquisition. and so, yeah, so that was a light bulb moment for us and that really just came across, from having a few of these, these customer interviews. Before sales. Like not actually reaching out and saying, Hey, we're gonna do sales. Reaching out and saying, we just wanna learn from your experience. Can we have some simple conversations? and people just opened up and and told us, not what we wanted to hear, cause we really wanted to hear. Yeah. What you're doing now is great for us. We'd love to buy that. Please. but what we needed to hear, and it, yeah, it transformed the trajectory of the company. I love that. And just to frame and put it on a wall, this piece of advice, if you can get the right customers in a room and not talk about the solution, but understand their problem, you will hear the language that they use and it might connect, it might not be the same language you're using, but understanding the language your customer uses to talk about their problem is the gift that keeps on giving. That's it. It's what? So Amy Jo, what you taught me is sometimes it's better to have a, your scientist hat on, than your salesperson hat on. So sometimes founders find it very hard to talk about their company without pitching. Right. And so, like the antidote to that is, stopping talking about your company, asking about your customer, asking about their problems day to day, and then, shutting up and listening. Right. And listening is just so, so powerful. cause at some point, even though most people can't get a word in edgewise when they're talking to me at some point they'll, you know, they'll actually explain what they're struggling with and it will make so much sense. Awesome. Thank you so much, Robin, for leveling up our thinking and giving us some really , exciting and challenging visions to wrap our minds around. And fun. I think Amy, Joe, like there is a lot of fun, uh, ahead of us as well. Right. So it's super cool to have the power of computation in the hands of everybody. Right. I think it's, um, it's a it could be a very fun future. It is. And I'm working with more and more AI startups and I'll be sharing that for all of you over the next few months, so thank you so much for joining us. Yeah, thanks, uh, for inviting me and having me here. it was just amazing. Thanks everyone. Let's get smarter together. There's nothing better. Bye.