Noah Brier is the founder of BrXnD.ai which he describes as ‘an organization at the intersection of marketing and AI.’ Mostly, I think it has been relatively small events where super smart marketing folks experiment about the impact of AI on marketing. Definitely take a look at CollXbs - a brand collaboration engine. And here’s a recap of the BrXnD NYC 2024 Event.
I first encountered Noah way back in 2008 when he launched Brand Tags which was like a free association tool for brands. Before that, he was the Co-Founder and CEO of Percolate, and the Head of Planning and Strategy at Barbarian Group.
AI Summary.
I start all these interviews with the same question that I borrowed from a friend of mine. She's an oral historian; she helps people tell their story. I love the question, but it's a big question. So I always over explain it. And so to say, before I ask it, you're in absolute control. You can answer or not answer any way that you want to. And the question is, where do you come from?
Where do I come from? Yeah, I'll give the sort of biographical details and then I will give the sort of real answer because I was actually at a conference, a small conference a friend of mine puts on a few years ago. And a CMO was having a conversation. He was saying that one of the things he's come to realize is that like when he's doing interviews, he is not particularly interested in sort of people giving their, he asked what's your story, right? Like the typical sort of interview question. People are, they go through and they're like, they graduated high school and they studied this in college and they went through often he's interviewing people who have just graduated college and they don't have much of a story. And he's much more interested in the story of how they are, who they are.
So the very fast biographical version of it is that I started my career actually very beginning as a journalist. I moved into the advertising industry, taught myself to write code. I started Percolate, my first software company in 2011. And I am currently doing this thing called brand, which I still have not figured out the perfect way to describe, but it is a, I call it an organization at the intersection of marketing and AI, and it is a house for all of the different activities that I have going on at the moment. At that intersection, whether it's a conference or consulting or building things or doing whatever I think probably the more interesting version of the answer is I am a curious person and a researcher.
There is a story from when I was in 7th grade, I think it was 7th grade, might have been 8th grade, I think it was 7th though where every December, my history teacher would give the entire class a question to answer over Christmas break. And if you could come back with the answer and the source material where you found it, then you would get an automatic A for the next quarter. You didn't have to take a test. You didn't have to do anything. So obviously when you're 12, that's pretty appealing concept.
So I went straight to work and I started doing all this research and I had my parents take me, I grew up in Connecticut. They took me to these libraries all over the state where I could find different things. And the question was about this diplomatic situation that happened between the U.S. and France called the XYZ affair, which was an attempt to, if I remember my history correctly, the U.S. was upset with the French or the French were upset with the U.S. over something around the American Revolution and having to do with the sort of relationship with the British. And the French were I think sinking U.S. ships somewhere, merchant ships. And I might be totally wrong about history, but this is not the important part. So anyway, U.S. sent a sort of a group of folks to go negotiate with the French and the French sent some sort of emissaries who came to go by X, Y, and Z. And they went by X, Y, and Z because they ended up offering bribes. Or they ended up requesting bribes basically. And so the question was famously it's called the XYZ affair, and it was three men, X, Y, and Z, and the question was, who was the woman that accompanied the three men for this negotiation?
And you had to go find it in this history books. And so after a few days of research, I found it. I found the answer. And I found the source. And then for some reason, along the way, I was like, it's really weird. I found sort of two references to her, but then every book published after 1983 didn't have any reference to this woman. And I was like, that's strange. And I don't really understand why that would be. That doesn't make a lot of sense. And I don't know. I just had a weird feeling about it. And you're a researcher about that sort of weird feeling where you're like, there's something that like, doesn't fit quite right here. And actually that's like the best feeling when you're like, I think I found something right.
And so for some reason, which I still don't remember, I decided that I was going to reach out to the author of this book. That was the turning point where after he published his book, it was no longer mentioned that she was there. And I was like I'm 12. And so I'm like, I'm going to call his office. He was a professor at Syracuse University. His name was William Stinchcomb, I believe. And I was like, I'll call, but then I'll just leave a message and I'll ask him to email me the answer. And I call and I get somebody else answers the phone. I asked to be connected to his voicemail and they just connect me to him.
And so here I am, I'm 12, I'm connected to this professor. And I stumbled my way through the story of my teacher and asking this question. He was like, you've asked the right person because I'm the preeminent scholar on the XYZ affair. And you're correct that my book does not mention that woman because I am the person who proved that she was not there. That there, this was a mistake. And your teacher is wrong. He had been asking this question for 15 years. He's been wrong the entire time he's been asking that question. That sort of information is out of date. And here's where you can find it. And in fact, the Yale library should have the copy of the Marshall papers that you'll be able to find this sort of reference in. And then I was like, can you also send me an email that says all this so I could show it to my teacher because he's never going to believe this happened. And he was like, sure.
So I got this email. And I walked in and my parents took me to the Yale library and I found the Marshall papers that he referenced. And I walked into my teacher and there was one person in front of me. So the rule was whoever got there first, got the A, and the woman in front of me or then 12-year-old in front of me had the answer. This woman, her name was Madam de Villette. And so that was it. It seemed like it was over. And I was like I actually have a different answer to this question. And so then I presented my research and it turned out and he took it well. I got my A. So did she, actually got her A too. I didn't have to do any work and then he retired the question.
But to me, to answer the question you originally asked, how did I get here? I think that I'm just an addict for that feeling where you have these things and you want to find the way they uniquely connect. And that is probably my favorite thing to do in the whole world. And I think a part of it is that I sort of came early to that particular feeling.
You tell that story. So you were 12, then what, do you have a memory of what you wanted to be when you grew up? Did you know, or have an idea?
No, not really. I didn't, I may have, I'm sure I had some sense of something, but to be honest, I'm still not totally sure what I want to do or be. I think I've been pretty lucky throughout my whole life and both educationally and professionally to be able to just follow my nose. I even went to NYU, but I was in a college at NYU called Gallatin, and at Gallatin, you got to rewrite your major every semester. And so that's just been my way of being, is do whatever is interesting, and taught myself to write code, and then I started a software company, and now I'm playing with AI and I've started this AI-ish thing. So no, I doubt I knew and I still am not totally sure I do. But at the moment I'm lucky enough to be at a point in my career where I can optimize for having a good time and still make enough money to support myself and my family. So it's a good situation.
What do you love about what you're doing right now? Like where's the joy in what you have going on?
Yeah, I think it's a couple things. For one, I spent almost a decade running a software company. And that was an amazing experience. I built the first version of Percolate, the product. And I was the CEO, but at some point that got to be so big that if you're asking any sort of CEO of a, even small company, you spend less and less time doing the fun, interesting things, and you spend more and more time dealing with interpersonal issues between people and figuring out which health insurance plan the company is going to be on next year. And so part of the joy in what I'm doing is that I get to make stuff all the time. And that is probably the thing that I enjoy most, whether it's writing or putting on a conference or writing code and just like seeing ideas come into the world is something I really enjoy. And I do think it fits into that story. I like to fill those gaps. I like to have that feeling that you've identified somewhere in the middle and then to actually make it happen.
I think that's a big piece of it. I'd say the AI piece specifically is just, this is the most amazing technology I've personally ever experienced. I remember getting my first computer, but I was not someone who was super aware of what life was like before computers. I didn't live in a world before computers, but it wasn't something that was super apparent to me. And I remember getting on the internet and I was building websites early, but even then I'm still relatively young and it seemed like that's just how the world worked a little bit. And this is just the most magical, strange piece of tech that I've ever used. I feel like I'm living through something that is just amazing. And I think a lot of people feel very uncomfortable in those situations. And I think it is uncomfortable. It brings into question lots of really fundamental stuff about what makes us human and what creativity is and all of these pieces. But it's also just really cool. I don't know, every day there's a world of things I can do today that I couldn't do two years ago. And that's just an amazing feeling if what you like to do most is make stuff.
My first interaction with you was Brand Tags. And then Percolate, and now BrXnD. You show up, it seems to me, in service of brand at these technological shifts. And I just wonder, when did you first encounter brand and what made it so interesting to you?
My first job out of college, I wrote for a magazine called American Demographics. It was a trade publication. Not a particularly successful one because to be a successful trade publication, you need to be laser focused on a single audience. And this one was focused across marketers and demographers and a couple other groups of people.
But I got this really amazing opportunity there where I met the editor and I pitched him a story while I was still in my senior year about Shepard Fairey and Obey Giant as a sort of interesting lens into brand building. And I was a big fan of graffiti and had been for a long time and I've been tracking Shepard Fairey and this was 2003. So I pitched it and he basically said, okay, if you write it, and it's good enough, I'll give you a job. And so I took a shot at it and it was an amazing thing to me. I'm sitting in my dorm room and I'm on the phone with Shepard Fairey and I'm on the phone with the artist and like all of these people who I had grown up just idolizing. I still think graffiti as a sort of art form is absolutely amazing. And I wrote this story about the idea of building this sort of guerrilla approach to brand building. And I think part of where that connection came from was in college, I ended up studying media technology and culture. And I was very focused on, I'm a big Marshall McLuhan fan. And so I think the sort of brand side of things came naturally.
I was interested in this idea of these things that are such kind of strange cultural artifacts. They don't really exist, but they exist in our heads and they live with these visual representations, but the visual representations are just a sort of tiny piece of this much larger puzzle about what they are. And so I pitched that story and he took it and I got a job and that was my first job. And from there I ended up because the magazine was focused around marketing and technology. I ended up talking to a lot of folks at that intersection. And when the magazine went out of business fairly soon after I got hired, because like I said, it was not a particularly successful magazine and I needed to get a new job because we all got laid off, I reached out to honestly, just a bunch of the folks who I had interviewed for these various stories and those tended to be either from marketing or tech. And so I ended up as a copywriter at an agency. Yeah, so I think it just came naturally. And then all these other things, I think I've just continued to be fascinated by brands and how they work and what they are. And they're very strange and unique nature.
And the fact that brand tags, which you mentioned, I did in 2008, I think. And it was this sort of experiment to understand what a brand was. And it was inspired by this article my friend Martin Bihl had written that basically argued that brands live in people's heads. And I thought that was a really interesting way to think about it that I had not thought before. And I made this thing that flashed up a logo and people typed in the first thing that popped into their head and made a tag cloud out of the results. And so it was a way to capture that perception in people's heads. And I think I've been just hooked on those ideas since then.
And I honestly, I learned to write code in 2010 to build that thing. I learned to write code to build brand tags. I had the idea and then I was like, I got to figure out how to do this. And so it was in service of this stuff. And then I think I've just naturally kept going with it. And then at some point in your career, you settle into this area where you're surrounded by people who are thinking about these kinds of things. But I also think it's more broad. I just think brands are these very important, unique things in this kind of strange ecosystem of commerce and culture and I continue to find them interesting all these years later.
The second BrXnD conference was just last week. Congratulations. And thank you. And congratulations. How are you feeling about the conference last week?
It was great. I thought I was on stage, so I'd be more interested to hear how you thought of it from the audience. Hopefully the chairs were more comfortable than they were last year. No, I thought it went excellent. It was fun to revisit everything a year later. It was really fun to have an audience of so many people who were returning. I thought that was really cool and unique. And I think that as much as things have changed over the last 12 months in this world of AI, we haven't progressed all that much. I think we're still in such early phases of this technology, even though it's quite old in theory. Neural networks have been around for a really long time, transformers have been around for, I don't know, six or seven years now. But we're just at the beginning of its impact on the world. And I think part of what I was hoping to do with the day last year was just demystify it. And I think this year, it's continuing that and pushing people away from feeling too certain that they are ready to declare mission accomplished. We know what it is. It's done. And I just think we aren't going to possibly be able to see all of the downstream impacts of this thing for many years to come.
The talk I gave to open the day I drew this analogy between bicycles, which was multi-layered. I first started using the bicycle analogy last year as a way to say AI is like a bicycle in that you have to get on it to learn it. You can't read about it. Bicycles have this sort of incredibly counterintuitive physics where you need to turn right in order to turn left and turn left in order to turn right. And that's the reason that we have to put a six-year-old on a bicycle and run alongside them, not read them a book about the physics of bicycles and then have them get going. And I think AI works very similarly. It's just a little bit too weird and counterintuitive for us to read about and understand. And so all we're left with is to play and tinker and explore.
But then as you look deeper at the bicycle analogy, which is something I started to do, there are all these other sort of very interesting layers, right? The bicycle was invented in the early 1800s, but it wasn't until the introduction of the safety bicycle, which gave bicycles two equal size wheels, that it became popular. And then you had this bicycle craze and people were very worried about it and what it was going to do to culture and what was going to happen. Interestingly, people were particularly worried about women and their ability to now move freely about what would that do to culture? And then if you want to continue to extend the analogy, which is something that I did in the talk, you can look at all these sort of second and third order effects.
One of them is that it did give women far more freedom and independence. Another is that it has this linkage between airplanes. The Wright brothers were bicycle mechanics and bike builders before, and learned to use high strength, low weight materials, because they were bike builders. And it has this linkage to cars because it gave us this view into what was possible with independent transportation. We had the steam railroad, but bicycles inspired people to use the roads in new ways and to push towards self-propelling technology.
And then the last sort of interesting analogy is that Steve Jobs read this article in the early seventies from Scientific American about how amazing bicycle technology was and all these linkages, and it inspired him to call the first PC a bicycle of the mind and talk of people as tool builders. And so I just think it's like, if you would ask somebody in 1869, when the New York Times was writing about the bicycle craze, what was to come from bicycles? They probably would have given you this very sort of linear story when you zoom out enough, what is airplanes and cars and computers and all sorts of crazy stuff that would have been not just impossible to imagine at that time, but silly to even project.
Yeah, I really enjoyed the events. I've just really appreciated how you contextualize all of it in a way that makes it very accessible. The idea of the deeply counterintuitive way that generative AI has shown up in our world. You guys talked about it and I can't tell if I'm making too big a deal of it, but thinking about how they launched generative AI. I rode the train down with a guy I know. Our children are friends and he was telling me a story about his experience with generative AI basically as a search engine - as a search engine that doesn't work, because that was the expectation he was given.
You’ve said, I think, this is one of the worst branding example ever. Launching generative AI into the world in the context of search, which sets up the sort of information retrieval expectations. And Tim Hwang, last year, points out that it's a Concept Retrieval System, not an Information Retrieval System. But it was introduced into the world as an information retrieval system. And so how many people are using it with the wrong expectations and are just... I saw last week OpenAI is launching a search engine, and it felt like an Onion headline, honestly. So I wonder, what do you think the actual consequences have been for generative AI that nobody took responsibility for setting the appropriate expectations?
First off, I will say that OpenAI introduction is in two and a half hours from now. We'll see exactly what it is. And Sam Altman has denied that it's a search engine.
Wow.
But yeah, I'm not sure that it's, I think this is the way technology cycles through, particularly when it's weird. I think it was a strange decision on the part of the technology companies to decide to apply it first to search, because it is a place where you are looking for a very specific answer. And this is not the thing it's best at. One of the things that I've come to see and talk about with folks over the last 12 months is this idea that hallucinations are a feature of these things, not just a feature, they're fundamental. There is no model without hallucinations. That's all they do. The fact that they're right so often is what should surprise us, not the fact that they're wrong, right? Because there is no sort of information retrieval, as Tim said, it's all sort of concept retrieval. And so the fact that they pack so many concepts in there and that they so often are correct, like technically correct, is a kind of amazing thing. Much more amazing than when they're not correct. But naturally, I think the easiest places to apply it are places where you can lean into that, not away from it. Where the goal is to lean into the fact that it's all a hallucination, not to try to fix it at all costs. And I think that's why creative projects are the most interesting. I don't know if you've seen this thing called WebSim, which is the most interesting and amazing thing I've seen from somebody over the last few months. Somebody built basically a browser. It's a fake browser and all it is just an AI imagined site. So you prompt it and it sends you to this fake website that the AI has imagined. And it's wild. It's insane. It's amazing. And that is magic, right? That's the magic of this thing. And when you lean into it fully, you get these amazing results. And when you try to fix it, but I guess if I zoomed out a little more, I'd say the technology companies are just making the same mistake that humans are in that, like, when you see something that looks like a computer, I made a joke at the conference that AI doesn't pass the duck test. The duck test is if it quacks like a duck and it walks like a duck and it flies like a duck, then it's probably a duck, right? And so AI looks like a computer. It talks like a computer. It is shaped like a computer. In fact, we experience it through a computer, right? We experience it through software, like deterministic software, deterministically written software. And so I think it's not a huge surprise that we're a little bit confused about what it is and that we ask it to do math, for instance, right? Because every computer can do math easily and they can multiply five times six numbers because it's a deterministic process and it's core to how it works. And not literally, but I think what we're seeing is a lot of people going to ChatGPT and figuratively asking it to do math, which could also be fact retrieval or any of these other things that we're very used to a computer being able to do perfectly. And then it can't do it and they're like, this computer is broken. And they're not wrong. The flip side is if five years ago, I was like, hey, Peter, can you write me a, have your computer write me a sonnet? You'd just look at me cross eyed. That's a silly idea. It's not even something that your computer can't do, it's something that you'd look at me and think I think maybe this person needs to be institutionalized or something. I don't, what is he talking about? Asking me to have my computer write a sonnet. And I think that's what we're all doing. We're all groping and grasping for whatever we can here. And we're trying to figure it out. And I do think it's strange that they've all chosen to start with search engines. And I don't think the search engine results are particularly good. And I've tried to use these AI search engines, like Perplexity. And I have to say I don't think they're great. I'm into just playing with stuff and trying it out and seeing what works. But I do think if you said to me, which is the more interesting project, like an AI search engine or this WebSim thing where people are imagining a whole world of stuff that's never existed before and will never exist, it's absolutely the WebSim stuff. Because we're doing something with this technology that we couldn't do before.
What was your first experience with generative AI that kind of made you, what's the right word, take stock, where you really felt like you had encountered something totally new?
I had been using GitHub Copilot for a while, which I think was probably my first sort of deep experience with it. If I rewind all the way, Andrej Karpathy wrote a post in, I don't know, 2011 or 12 called the, maybe it was a little later, called the unreasonable effectiveness of RNNs, recurrent neural networks. And in it, he had instructions on how to train your own RNN. And I had done that, whatever, in 2012 or 13 and just played with it. And it was crappy. And I tried to make a thing that could talk like McLuhan. And I had some fun ideas. In fact, the idea I wanted to build out of that, because those RNNs were not very good and you needed to train them on a huge corpus of materials in order to make them work. But one of the interesting things that kept coming out of it was these misspellings and these funny, close, but not quite right words. And I wanted to make a dictionary of impossible words. For me, that was my idea back then. And so that was my first experience really playing with it. And then GitHub Copilot was probably my first regular thing. I was building a lot. And what I found particularly interesting about that is I was writing a lot of tests for product that I was working on. And what was amazing is that once you had enough of these tests in your code base, basically I could just write the name of a new test and it would just pop the test out. And I thought, huh, that's really interesting. But then the first sort of real thing was I discovered that this was GPT-3, and I discovered that you could give it a data structure that you wanted data returned in, and it could return structured data from unstructured data. And that was the first thing where I was like, wow, I am never going to do this any other way ever again. This is so far superior. I'd done a lot of web scraping in my life and web scraping is this terrible process where you write this very brittle code that says here's the title tag and here's the H1 tag and here's the paragraph and go grab those pieces. And if one tiny little thing changes on the site, everything breaks. And here with AI, I could just grab the text, give it to the AI along with a structure and say, hey, parse out all the pricing information and give me a CSV with each row as a plan and the plan has a name and a description and a price and the prices per whatever. And here are the features. And that was mind blowing because it was like, I think those moments where you realize, oh, I'm never going to do this any other way ever again. And I wouldn't say actually I have had a ton of other, that is still the number one thing I will never do any other way is structured unstructured data. But that is a continually useful thing that I need almost every day in everything I do. That is probably my number one use case at its broadest.
I think the first thing you did, I'm not sure the chronology, but you did the BrXnD COLLXB, right? And so you had an AI create these collaborations between brands. And one of the lessons or findings was that when you asked it to make a sneaker, it put a Nike logo on it, right? And it was almost like the evidence of brand in LLMs. I'm not even sure how to talk about it. The evidence of brand equity in LLMs. Is that, was that the right way of saying it?
Yeah, I think that's reasonable. Yeah, so I built this thing. You could smash any two brands together. It's still available at brand.ai, B R X N D.ai. And one of the things I kept seeing is that if you ask for different brand sneakers, collabs, even if it wasn't a Nike, it would often come out with a swoosh on it. And what I thought was interesting about that was that's technically a hallucination, right? It's technically incorrect. But my argument was that it's perceptually correct, right? That is the way people think about sneakers. If we went and surveyed a thousand people and we said what logo goes on a sneaker, a huge portion of those would say, if you asked them to draw it, a huge portion would put a swoosh on it, right? Because to many people, that's just a sneaker. It's not a Nike, it's not a brand even, right? It's transcendent. And I would say that lesson actually has continually repeated itself. I'm working on a project now with a large radio company, I'll say. And one of the things that we've been going through is, the real challenge is getting these things to be less technically correct, like it comes back to that hallucination thing. In my day to day work, I generally don't find that the challenge is that these things hallucinate too much. It's that they are too specifically correct. I sometimes equate it to it's like working with the most junior employee who does exactly the thing you said. And you're like, but that's not really what I meant. I expected that you would interpret that I need you to spend five more minutes on that. And they come back after five more minutes and you're like, no, I meant I needed it to be better. That's what the AI does. It takes you literally, but not always correctly. And so it's again, in this sort of world where everybody's so focused on hallucinations are bad. Again, it's just not, it's not really a big part of my experience. I find it to be the opposite. I find that again, it's like working with those junior employees where the frustrating part is not that they don't listen to you. It's that they don't listen to themselves, that they don't have that experience and internal understanding. And so they take you too literally and you think in your mind, it was obvious that you were looking for, hey, I just need this better. I have a ongoing joke about the number of times that a CMO or some other leader says, hey, have you thought of putting that in red? And then everything comes back red from the agency. And they're like, no guys, I was literally asking you, did you think about that? Like you could have said no, and that would have been fine. I was not telling you to bring it back in red. So I feel like that's a constant source.
You mentioned consulting and stuff. I'm just curious, what kind of questions are people coming to you to answer? And what is brand management and brand building mean in this new era?
I'm getting asked a couple of different classes of questions, I'd say. One of them is just at a high level. Like, how do we adopt and integrate this technology into our organization? I'd say that's probably the most common one. We don't know exactly what we want to do with it, but we know that we should learn more about it and we should have a point of view on where it belongs. And how do we develop that point of view? And then there's more specific ones, which are like, hey, we have some ideas. We need some help on figuring out how to make it happen, whether that's help with prompting or building specific things. And then once in a while, I do a specific kind of build or project. I tend not to build a lot for other people, to be honest, just cause it's not really the thing that I think I'm best suited to do. My skill is I think as a prototyper, not as a sort of production level software developer. I would take my ability to sit inside an organization and build a prototype of something against almost anyone's. I know my own limitations well enough to know that I'm not the person who should be putting this stuff into production and making sure it's got all of the security and all the different pieces that it needs. That's throughout my career, that's where I've hired much more talented software developers than myself to do that job.
I know you just coming off the second of the conferences, but what's next for BrXnD.AI? And what are you most excited about?
I've got lots of stuff going on. So I have a whole bunch of clients that I work with on an ongoing basis on these kinds of questions. How do I adopt and integrate AI? I've got to turn around and be on the road in two weeks with 30 execs to put on a four hour version of the conference. So that's one thing. And I've got ongoing bits and pieces there. I have some ideas about doing more verticalized versions of the conference. I think there's some specific areas I'd like to dig in on more, legal is one of them. I think potentially there's some interesting things to do with creative. So that's one approach. I think I'm personally struggling with this question of I love this 200 person conference, it feels intimate enough that you can really do fun things. And I don't want to break it. I think I want to keep it that size, but I also have a bunch of other people and ideas who would like to come. So I think that might be one approach is to make it much more specific. And then I just have a never ending supply of bits and pieces and little projects and things that I'm working on, I have an ever-expanding personal assistant project that I've been building that has AI components, just this kind of amorphous code base I've been building that can do all this stuff for me whenever I need it. And that's been fun whenever I think of something, I send it that way.
How is that going?
It's good. I don't know, whatever I use it. Some of, a lot of this stuff is not particularly complicated or even particularly AI-ish.
And to be clear, you're building an assistant for yourself, is that what you're saying?
For myself, yeah. So I can email it or text it or do a bunch of other things and it's not a super smart agent or anything that can do a kind of endless supply of things, but it's something where I keep layering on these specific tasks that it can complete. Again, it's a mix of AI stuff and non AI stuff. But it's a fun ongoing project and every time I think of something that I'm like, I wish that I had this thing, I just do it and whether it's one of the functions is it takes my receipts and pulls out everything and files it away for me or another one is that when I have a link that needs to go somewhere it'll go scrape it and summarize it and do all these things. So I have this set of functions. So that's a project I'm always working on a little bit. I'm pretty interested in finding a good approach to building a retrieval-augmented generation app for myself. Maybe that's something I'll integrate into that assistant. But I want somewhere where all of my writing is and I can access it and use it for various purposes. So just fun, weird projects.
Nice. I really appreciate you taking the time and yeah, I enjoyed the conferences. It's been fun talking to you. So thank you very much.
Yeah. Thanks, man.
