Subscribe
Share
Share
Embed
Built This Week is a weekly podcast where real builders share what they're shipping, the AI tools they're trying, and the tech news that actually matters. Hosted by Sam and Jordan from Ryz Labs, the show offers a raw, inside look at building products in the AI era—no fluff, no performative hype, just honest takes and practical insights from the front lines.
Wanted to build a trading bot leveraging your API. I've got it in dark mode, light mode, and then my favorite mode, disco mode.
Sam Nadler:Party mode.
Sam Nadler:Hey, everyone, and welcome to another episode of Built This Week, the podcast where we share what we're building, how we're building it, and how it affects the world of AI and startups. I'm Sam Nadler, cofounder of Ryz Labs, and I'm joined every week by my friend, business partner, and cohost Jordan Metzner. And this week, we have two special guests, Jake and Emily from Product at Public. Please introduce yourself, Jake and Emily, and then we'll get into the tool we built this week, the public tool, and then also some hot AI news.
Emily Kurtz:Awesome. I'll get started. Thanks for having us. I'm Emily, head of Product at Public, and have been really responsible for all of our trading. So the API, which we'll talk about shortly, Belton can really speak to.
Jake Trefethen:I'm Jake. Thanks for having us, Sam. I lead AI at Public, all of our generative AI offerings, Alpha, kinda throughout the app. So thanks for having us. Excited to get started.
Jake Trefethen:Great.
Sam Nadler:Awesome. Well, Jordan, like, every week, we we try and highlight a tool we built in the last week. This week uses the public API. Do you wanna walk us through, a, just what you were thinking about before the build, you know, what the tool does, and then lastly, how you built it? And we have, you know, obviously, the team, the experts here, Emily and Jake, to give us feedback and maybe point out things we could've done better or things we missed or just overall how we did.
Jordan Metzner:Cool. So maybe just a little background, at least from my perspective. I am no expert stock trader. So, you know, my expertise, in trading stocks is kind of my own personal, portfolio. I'm not a day trader.
Jordan Metzner:I don't really spend my time trading stocks on a daily basis, but I think kind of that lens to just how easy it was for me to leverage the public API to get to get an app working. And I love to tell you guys just a little bit about that. I mean, it just took me a few days to get the entire app working end to end. But maybe before I jump into that, Emily, can you just maybe tell people a little bit about, like, what is Public?
Emily Kurtz:Yep. Of course. So Public is a brokerage that makes you know, is for those who invest and take it seriously. We offer the whole suite of asset classes, stocks, bonds, treasuries, options, crypto, as well as unique yield products like bond account, treasury account, and HEICO. And so you can open brokerage accounts, Roth IRAs, IRAs, and more.
Jordan Metzner:Okay. Cool. And then maybe a little bit about kind of API and, you know, what its use cases are and whatnot.
Emily Kurtz:Of course. So we recently launched our API. It's really to think of it as like fourth front end for people who don't necessarily want to use our UI, but really want execution capabilities and the ability to programmatically access their account. We started sort of with our core capabilities, so think month trading options, trading equities. We are continuously building out that road map, adding on trading crypto, trading bonds, money movements, and then we're also adding technical functionalities.
Emily Kurtz:Right now, it's really a REST interface, but we are working on WebSockets, NCP, Python STKs, and more.
Jordan Metzner:Awesome. Okay. Cool. So I guess a little background. You know, I didn't really have, you know, too much to go on.
Jordan Metzner:I wanted to build a trading bot leveraging your API. API. So maybe I can just jump in to show you guys what I built, and you haven't seen this yet. So
Emily Kurtz:No. We're excited.
Jordan Metzner:Okay. So I was getting too aggressive, so I had to stop trading. But, essentially, the way the trading bot works is it leverages the public API as well as ChatGPT. And I built actually the entire app in Replit. I think total build time was probably somewhere between like three to six hours total.
Jordan Metzner:I vibe coded the whole thing, so I didn't all of the public API was pretty much available within, like, Replit's, like, knowledge base. When I wanted to do certain things, I kinda copy and pasted certain things from the public documentation. Let me just tell you how our system work. What it does, it reads the price of a certain stock. In this case, I chose ibit, which is the Bitcoin ETF.
Jordan Metzner:And then what it does is it sends that price plus how much money I have in my portfolio and how much risk I wanna take over to ChatGPT. And then ChatGPT actually does some analysis and decides whether or not I should buy or sell. And then once it does that, you can see here ChatGPT is making different confidence intervals. Here, I think it has every thirty seconds I have it running. So it'll go in ChatGPT, tell it about the price, tell it about my portfolio, tell it about how much units I have currently.
Jordan Metzner:I also have in my settings, I have a stop loss, stop loss percentage, so I make sure I never run out of money. I kinda have a percentage of max size position. I also can change my aggressivity levels, etcetera. And I have kind of a float available like in my algorithm. So if I don't like run out of I never can run out of money.
Jordan Metzner:I always have like a little extra cash on hand in case like the market goes like, you know, like really to the upside where I wanna make a a buying decision. And not all the pages like work perfectly, but most of them should work. I kinda did some like AI predictions, so I could I guess this one doesn't work perfectly, but so it's kinda predict like how well this will work. But I have kind of all the technical performance and live monitoring, and then here's my dashboard. And then here inside the public account so what you can see is, you know, essentially here it started to buy shares and sell shares and buy shares and sell shares, and it just kinda went a little bit crazy.
Jordan Metzner:So after it bought and sold four shares today, where it looks like I lost about 11¢ in the trade, I turned it off because we were gonna do the episode, so I didn't want it to go too crazy. And then I got two more little bits, so I built it all with Replit. That's actually doing all the front end and the back end. And then ChattyPT API, I've got it in dark mode, light mode, and then my favorite mode, disco mode. Party mode.
Jake Trefethen:I love this. Alright.
Jordan Metzner:What do you think?
Jake Trefethen:They should they should make disco mode a a default option on all on all browsers.
Jordan Metzner:Okay. On all web apps.
Sam Nadler:Yeah. Was the feature Jordan was most proud of.
Emily Kurtz:Well, Jordan, I love this. This is this is awesome. Obviously, you're pulling in from Public's APIs, portfolio orders, history, etcetera, but it also seems like you're piping in from other data sources that Public doesn't yet have available via APIs. So some of the potentially predictions, it sounds like maybe on the analytics side, you you have some stuff around its its ratings, performance, analyst reports, etcetera. So can you speak to sort of what else what else you're using?
Jordan Metzner:I didn't want to be overly aggressive in the rate limits on calling the API for pricing too aggressively. So I found a public API that was free that I could call, like, every fifteen seconds to get the price called Polygon. So I chose that to get the price. And if I run over my rate limit, I have backup as public as my backup. So I get the price about every fifteen seconds.
Jordan Metzner:And then we use ChatGPT APIs all the time. So I created a new API key with ChatGPT. I created a public API key as well. And then I created a prompt basically kind of going through ChatGPT, telling it where I where I was, what my aggressive levels were, what my portfolio was at. And since we're only trading one stock, it's pretty easy.
Jordan Metzner:I'm I'm sure like this would get infinitely harder as you started to trade more and more stocks. But in theory, you could paralyze it. So you could, you know, you could have a bot that only focuses on each stock and, you know, I don't know how good ChatGPT is at like making money. So it'd be fun to run it, like, for a few weeks to see, like, how well it does. But, you know and and especially, like, this stock is, like, pretty highly volatile.
Jordan Metzner:So, you know, probably, like, the more volatile it is, that might be, like, better or worse for the returns. And it's hard to say because I really don't know, and I'm not, like, a pro day trader or anything like that. And I just kinda let chat GPT do its thing. I I had the prompt ready, but I don't know where it is now. But, you know, from a high level, I just give it, like, what I have in my portfolio, you know, how much shares I have currently, you know, how aggressive I wanna be, my trading history, and then, like, how much I how much buying power I have and how much money I wanna save and, like, my extra float just in case kind of thing.
Emily Kurtz:That that's awesome. And just saying it out loud, obviously, this is not investment advice, but it's fun to be able to show something like this with a more volatile stock since you can really see and showcase algorithm.
Jordan Metzner:Totally not investment advice. This is not investment advice. I would not trade real I would not trade real money at scale using this algorithm, but the cool part is you could trade it against the history and you kind of see how well it would perform. It'd be good to at least inform traders on whether or not there'd be an opportunity to make a trade. And I think that's probably, like, the takeaway here is that, you know, you could use it to inform you to decide, like, hey.
Jordan Metzner:This is a good opportunity to come in or maybe a good opportunity to come out.
Emily Kurtz:Exactly. We talk a lot about trigger based trading. Right? When a price hits a high or a low, when it grows by x percent, and so all of those things are really helpful. Maybe an inflation report and economic report, etcetera.
Emily Kurtz:And so being able to sort of trigger trades based on things is exactly what you're showcasing and something a lot of our customers, you know, like and expect. Another question for you is, like, obviously, you built this with a bunch of REST APIs. How would this evolve, change if you had WebSockets for prices or WebSockets sockets for order updates?
Jordan Metzner:Yeah. So, I mean, the more streaming, the better, I think, because, like, then we can stream the price rather than call the price without rate limits and things like that. So it just allows the entire thing to be smoother and smoother. You know, one thing I think I was gonna ask Jake about, you know, I just chose ChatGPT, like, the newest API as, like, the API of choice, but I don't really know enough about does Gemini do a better job at kind of financial analysis for trading? Or does, you know, does Claude, you know, Opus or Sonnet have, like, better analysis for trading?
Jordan Metzner:So that might be another, like, weighted thing to do would be is, like, see if you could have it run against multiple models and get their feedback and try to do some consensus. I think if I was able to stream, I'd be able to do, like, more trades even faster. I also presume the cost of trading is zero. And, you know, it it is, but, you know, every time you're in a trade and out of trade, there is, like, some time that it does take even though, obviously, it's pretty quick. But every trade is an opportunity cost against another trade.
Jordan Metzner:Right? So if you're trading in one thing and then something happens somewhere else, and so, you know, again, going back to the simplicity of, like, having only one stock is really easy to follow. When when you have a more diverse portfolio, the complexity scales, obviously. Anybody can do this, you know, just this was Replit, just talking to Replit. I probably use ChatchipPT maybe to write a PRD or something like that to write a product requirement stock, but, yeah, it's stored on my API keys.
Jake Trefethen:No. Yeah. It's it's really interesting. I guess, curious to hear whether you tried out any kind of multimodal prompts with ChatGPT. So, you know, did like, sending a picture of the of the price chart versus sending the metrics.
Jake Trefethen:Yeah. So, you know, in experiments with this that I've done, something that's really interesting on your question around the model is if you use a multimodal model, you can actually send it a picture of the price chart and even multiple time frames of the price chart, and it does a really good job. The the models have improved a lot over the last couple years, but it does a super good job actually of doing, like, visual price action analysis. Interesting. Like, you can even overlay indicators and and whatnot, and it and it can actually process pretty efficiently.
Jake Trefethen:And so, yeah, like, that combined with something like a like a more agentic model, like a SONNET three five or like SONNET four, like, thinking type model that kinda Mhmm. Puts a lot of tokens behind its decisions is kinda would be kinda interesting for sure. As, like, an easy next step. Well, that's that's a trade off. Right?
Jordan Metzner:Like, if you wanna give the model more time to think, then you trade off kind of how fast you can run trades. So Yeah. You know, it really would depend on your strategy of how, like, frequently you wanna trade. I tried to build something that was, high frequency, I guess, the sense of, like, for me, every fifteen, thirty seconds is pretty high frequency. But, yeah, if we were just trading once a day or even, you know, every hour or something like that, you know, you might be wanna give a model, you know, five, ten, fifteen minutes to make a decision and to really, like, think deeply.
Jordan Metzner:Yeah. One question about the charts. Does it work the same if you give it, like, the JSON based history of the price action? Because one thing I did was the my first build, I just gave it, like, how much money I had. And then I started to improve it by giving it my trade history, which kinda gave it some influence of, like, you know, where I've bought it before.
Jordan Metzner:It doesn't have I didn't give it full price history, but to your point, I understand like, yeah, I
Jake Trefethen:guess it can understand like, you know, kind of the dips and valleys and whatnot. Yeah. Definitely giving it context of any kind helps. Right? Like, definitely, when I played with this, it had a lot of trouble with time frames.
Jake Trefethen:So dialing in and out of, you know, are we on a two minute here or two hour? But, yeah, I mean, giving it any historic context is definitely helpful. And, I mean, I think I saw something today that one of the models I think maybe it's a Claude model is now like a million token context window. So it they they can handle us an insane amount of historic context, multiple time frames all in one shot now. So something to think about there too.
Jake Trefethen:Loading up
Jordan Metzner:Yeah.
Jake Trefethen:Pretty much every time frame.
Jordan Metzner:Yeah. To your point, yeah. I wonder like how much more data I don't know where the diminishing returns are, but how much more data you can give it on the price behavior to get more insight, you know, depending on your trade strategy as well. Right?
Jake Trefethen:Yeah. I mean, I've heard like a quick rule of thumb is somewhere around 75% of the stated context windows when you get degradation. But just for like frame of reference, the Great Gatsby is like 60,000 tokens. So a million is a lot. I wonder if I
Jordan Metzner:could just do like a Great Gatsby trade strategy, you know. Yeah. Okay. Cool. Any other questions maybe from Sam, Emily, Jake about kinda what I built here?
Emily Kurtz:Only question for me is obviously you built this in three to six hours, like what were the things that you'd want to add on if you had a little bit more time or dive deeper into?
Jordan Metzner:Yeah. I mean, I think, like, in one way, like, it's really easy to make a trading, you know, a trading bot or, like, leverage the public API to, you know, build a web app. If I had more time, I would have probably built, like, a back end, like, probably connected it to Supabase and built, like, my own React front end and self hosted it and, like, been able to run some cron jobs, like, probably a little bit better. One thing I learned is that I don't know when the market open and closes. I mean, I know on a daily basis when the market open and closes, but there's actually holidays and, like, Labor Day is coming up, for example.
Jordan Metzner:So market time would be a good API that I learned because otherwise, I have to, like, kinda figure out if the market's open or if, like, figure out through another format or just presume it's always open unless it's not type of thing. So that was something I learned, like, thinking about. Like, I didn't really think about it. I just thought it's always open the same time, and then you stop thinking, oh, no. It closes on the holidays.
Jordan Metzner:And so
Emily Kurtz:Different asset classes have different market hours. So the bond market actually is different than equity market Deposit time frames can be different on holidays. And so having an API to pull that all in and and handle that correctly is is a good idea.
Jordan Metzner:Oh, yeah. That'd be super cool. But, yeah, I didn't know the markets trade at different times in, like, after hours markets and, like, can I I didn't know if I could trade after hours or not, but, like, you know, maybe I can for a little bit of period of time? So, yeah, that'd be cool to know.
Emily Kurtz:You can for for next steps.
Jordan Metzner:Okay. Cool. Alright. Well, anyway, I don't think I made much money. I don't think I lost much money.
Jordan Metzner:I think I did okay. Yeah.
Sam Nadler:What is what is the final outcome in in, I think, just a few hours? Are you up or down?
Jordan Metzner:I on the
Emily Kurtz:on your portfolio. Let's just see what that what that looks like. I think you started with $300, and you're at $2.99,
Jake Trefethen:so Yeah. Okay. We're we're making
Jordan Metzner:it this weekend. Alright. Not bad.
Jake Trefethen:Not so bad. Chad GBT didn't know it only got two round trip trades a day when it started. So it wasn't it wasn't optimizing that way.
Jordan Metzner:But if you look, I mean, one, two, three, four, five. I mean, I made a lot of trades here. Right? Ten, twelve trades or something like that. So when making that many trades, to say that I'm breakeven is actually pretty good, I would say.
Jordan Metzner:Totally. Especially because I don't know what I'm doing.
Sam Nadler:Okay. Great. Well, you know, thanks, Jordan. We did cover just, like, a brief intro about Public. But if you guys don't mind, I'd love to hear, like, kind of the wider perspective of of Public and Jake, like, how you're applying generative I AI and what your role entails.
Jake Trefethen:Yeah. For sure. I guess I can give a little quick intro on Public and then dial into the AI stuff. So, you know, we think about, like, public as kind of the home for modern investors. So, you know, I think generally the landscape has kind of two type of tech forward incumbents.
Jake Trefethen:You've got kind of robo advisers, and you've got self self directed kind of brokerages. And, you know, when you're looking at higher net worth individuals and more sophisticated investors, it's not always true that you want one or the other. You know, sometimes you want to be able to directly manage your accounts, and sometimes you want to have a little bit of help. And so, you know, we see kind of the modern, you know, the inevitability of this whole kind of platform shift to to round out to something where people can dial into their own preferred experience, you know, still have access to manage solutions, but also not have to, you know, go through a wealth advisor at Morgan Stanley just to initiate a a transfer. Right?
Jake Trefethen:And one of the big pillars we kind of see in what the modern, you know, sophisticated investing platform needs to have is a solid focus on on AI native features. I mean, we know that that's gonna be a huge part of the future. We've been kind of a first mover, I guess, amongst peers in that space. We launched Alpha, which is our investment research bot powered originally by GBT four just a couple months after, you know, Chad GBT released. We were one of the first Chad GBT plug ins.
Jake Trefethen:And then after we released Alpha, we introduced probably, like, six or seven different AI powered features across the brokerage experience. We launched one or two kind of external products as well that were, you know, vertically focused on AI. So we've done everything from real time, you know, event summarization, earnings call summaries
Emily Kurtz:Why is it moving?
Jake Trefethen:Portfolio construction, explaining, you know, why stocks, your portfolio market's moving. And we've we've kind of experimented a lot in the space, and we've we think we've gotten to a place now, I think, where we've identified what we see as the big value drivers for, you know, AI native, kind of the next phase of, like, you know, true value unlock in the space. So one of the big opportunities we see is obviously just summarization. There's so much unstructured content in this space between news cycles for stocks that you own, the market macro as a whole, and then, you know, just dialing into what's actually important for your portfolio and getting insights out of all of that kind of jargon that's, you know, spewing out of of the news cycle. And so we think AI has a real value add there.
Jake Trefethen:A second vertical we think is really interesting is assisting with portfolio construction, helping to kind of manage investments with the help of generative AI.
Emily Kurtz:And like an extension of that, right, is discovery. So how can we help you and they find assets that are relevant for you or that meet your needs versus your traditional, you know, search where you have to be searching for the exact symbols that you're looking for.
Jake Trefethen:Yeah. So as an example, we we started experimenting with this feature we're calling generated assets. But the idea is to let investors build custom indexes out of natural language kind of queries. So kind of imagine whatever strategy you wanna come up with, you can go back and forth with an AI and dial in kind of an LLM, agentic assisted index that, you know, eventually will hopefully be able to manage on an ongoing kind of basis. Imagine stuff like, you know, I wanna invest in GLP ones, or I wanna do an anti tariff hedge, or or something like that.
Jake Trefethen:And we see we see that kind of filling a gap in the in the space between your kind of traditional low cost ETFs, and then, you know, your kind of higher cost thematic ETFs, or even somewhere like a, you know, just like a personal wealth adviser. We think that's a really interesting space as well, and we're working a lot there.
Jordan Metzner:Yeah. Just, I guess, a little bit more maybe about Public and kind of what's its mission, and really what what what the goal of the company is.
Emily Kurtz:As I mentioned, Public is really a retail brokerage app, and our goal is really to status to to meet the needs of investors that take investing seriously. And so over the last year, we spent a lot of time on building that foundation, adding a variety of asset classes, adding fractional bonds, being the first to do that, adding, you know, a treasury ladder, really meeting the the market with yield products as well as also expanding on the more active tier front. So adding options, adding margin investing. And I think what we're really looking ex we're we're looking forward to over the next year is how we build upon that and really leverage AI to deliver deliver unique value propositions that we really feel like public can deliver in a way that the market has not seen yet and hasn't and hasn't had. And so I know Jake talked about generated assets.
Emily Kurtz:I think there's opportunities for us to create more portfolios across across asset classes, expand that into other account types. So rather than it just be your brokerage account, maybe using those in sort of trust entity other account types and really helping meet the needs of that investor that's looking to grow their wealth for the long term in a really current time frame way versus that, you know, historical traditional brokerage, where you have to call an adviser and and get advice.
Jordan Metzner:Wow. Awesome. Okay. Cool. Let's jump into it.
Jordan Metzner:I think that was great. We recorded last week the same day chat GPT five came out, so it's been almost a week now since it's came out, and it's been quite controversial for many different reasons. But, yeah, you know, Altman's obviously been pretty loud about trying to make changes quickly to satisfy the the public, but just curious, maybe to Sam, Jake, and Emily, kind of what have your been experiences on using ChatGPT five, and do you like it better or worse? I've heard mixed mixed opinions from pretty much everyone I know.
Sam Nadler:For me, it's been a little frustrating. There's been times, and I know this, like, auto fast thinking feature has come out, but there's been times when I've gotten the the quick answer when I want the thinking answer, and it's just I've had to, you know, re prompt to think deeply and just kinda it's there's been, I feel, more back and forth, whereas previously, I could just pick the model that I thought best suited my needs right away. What about you all, Jake and Emily? Have you had much experience with GPT five in the past six days?
Jake Trefethen:I think I definitely gotta echo some of your struggles with it. I don't think I've sent it a single prompt where it didn't try to just split test thinking mode and quick mode. So I don't know. It it sounded like from the article too that I don't know if the router was even really working when they launched, but it sounds like they've been dialing it in. I thought another interesting part from the article was that some customers kind of were mourning the loss of the GPT four o's personality, and that, you know, Altman acknowledged basically that, you know, we're gonna ultimately have to get to a point where managing the personality of your model is something that's on on like the customer side rather than on, like, a model selector kind of front, which is is pretty funny.
Jake Trefethen:And I think it said something also about, like, some funeral in San Francisco for Sonnet three five.
Emily Kurtz:I think it's the same. I think they really wanted, you know, to try to solve everyone's needs and sort of take the decisions out of your hands. And I think it's realizing that that was probably done a little prematurely for where the product was, and people are wanting that flexibility and and that control back.
Jordan Metzner:What about that it could be a cost savings mechanism where, basically, customers are using expensive models to do low quality tasks, and that's costing them a lot of money. And so, in fact, they wanna make the decision on which you know, how much resources to put towards the query, which will allow them to save a lot of money considering the scale.
Emily Kurtz:Yeah. I mean, I think it's a good point. I I I I'm supportive of that, especially when you think through the the impact on the environment of all of everything that's being run. But I think this ecosystem has been evolving so rapidly, and so customer, all of us just expect so much from them. And so, unfortunately, like, if you wanna make that decision, great, but it needs to be done in a product that's that's pretty robust and still doesn't feel like a step down to to all of our users, and I think, unfortunately, that's not what happened here.
Jake Trefethen:I was gonna say it's definitely hard for them to avoid the buildup of hype. I don't you know, it's been a while, people waiting on GPT five, and I think the allure of the name, you know, definitely kind of maybe got out of hand for them as far as, you know, what they were able to actually deliver. I know that I heard it conflated with, you know, that's AGI when we hit GPT five and and stuff like that for at least, you know, whatever, six months. So, you know, for it to be kind of a a router adjustment, cost saving, slightly different model is is always gonna be a disappointment regardless of of what broke on day one.
Jordan Metzner:I think I mean, it's disappointing. It doesn't it's not a better coding algorithm, and it didn't beat, you know, the Anthropic model, so seems like everyone's gotta get back to work.
Sam Nadler:I do wanna transition to the next. I do have three PMs on this call. Jordan, you are excited about this post. I don't I don't know what platform it was from, that Shopify is requiring vibe coding in all PM interview processes. Do you think that's fair?
Sam Nadler:Do you think that's necessary? Do you think that's ridiculous? What are your thoughts here?
Jordan Metzner:No. I mean, it seems like so obvious. How could you not ask, you know, a PM to vibe code in an interview? Just like how could you not ask a developer to use AI to code an interview now? And it almost be, you know, the equivalent of asking someone to do, you know, complicated math without a calculator just to see if they could, you know, do it by hand.
Jordan Metzner:And so, yeah, it seems like so obvious. So this seems like another opportunity space, but also just seems like, you know, the way of hiring and the way of building is changing so quickly, at least in, you know, what is our vertical, which is, you know, product development. And I think, like, everybody becomes a product manager, and then, like, product manager also becomes everybody. So don't know. There's a huge, like, opportunity to open up here.
Jordan Metzner:Curious what Emily's thoughts are, especially as, like, leading product teams.
Emily Kurtz:Totally. I mean, I think every day, right, our CEOs are saying how can everyone at the company use AI more efficiently to to job and and make it better. And so I think this is just, to your point, like a no brainer example. How can you do some of the work? How can you prototype?
Emily Kurtz:How can you put something in and do the the low hanging fruit to put something in front and then have the team and the folks that you hired to really code, do the more complex architectures side of the house.
Jordan Metzner:Yeah. What about you, Jake? Yeah. I'm interested to see how
Jake Trefethen:it evolves. I mean, obviously so, I mean, I have a technical background and a product manager, and so it's it's always, I guess, been helpful for me, especially, like, scoping AI features to understand, like, feasibility. So I think having a technical angle to any any PM is definitely helpful as we move into, you know, having to manage a lot more AI style kind of development. But I I think it'll be interesting to see how it evolves, you know, the the specific use of vibe coding from a PM. I think today, PM's vibe coded prototype, the the code's not gonna go so far as to maybe reach production.
Jake Trefethen:I think there may be some, you know, developer pushback, and I don't know how reusable it's all gonna be. I think, you know you know, Replit, Cursor, very good at using standard, like, UI component libraries, Shed CN, whatever. And so a lot of it ends up being throwaway code, but that can be very helpful in iterating through design cycles really quickly, not having to involve a bunch of different parties communication overhead and getting from idea to something that's executable super quickly with one person. I think that's that's where, like, the real time time value leverage is right now. But I wonder how it'll evolve, you know, into actually, you know, involving PMs and production code.
Emily Kurtz:Yeah. And I think to Jake's point, right, like, it also depends on what you're building, because we are a brokerage, and we're responsible for investing in folks' money. Obviously, like, making me vibe code something that we're gonna put into production that directly impacts, their withdrawals, funds, whatever it may be, is it's pretty risky. But, obviously, if it was if I was by putting a Slack integration around some product updates, that's a lot less risky. And so I think there is we still have to figure out the balance of when to use it, when it's just a prototype, and when it can really go live in deep production.
Sam Nadler:Yeah. I'm curious how much of a signal it really would be because, you know, I'm not a PM. I'm an operator. But, you know, in a thirty minute period, an hour period, I can vibe go to, at least in my opinion, a a decent prototype for almost anything. So, and I probably wouldn't hire myself as a PM.
Sam Nadler:So I'm just not sure how, how strong of a signal it would really be. But I would agree that PMs should be comfortable in creating quick prototypes for whatever they're working on.
Emily Kurtz:Totally. And and to your point, it's it's definitely not the only thing that's important. Right? How you work cross functionally, how you manage bunch of different stakeholders, how you prioritize, how you bring in the voice of the customer and communicate with them, how you make a team more efficient and sort of push them to be the best product velocity they can, all of those things you're not gonna get via coding.
Jake Trefethen:Exactly. To your point of a signal, hiring somebody who's supposed to be inspiring others to ship software products, if they haven't played around with a a vibe coding tool and tried to build something themselves, I'd definitely question their curiosity a little bit.
Jordan Metzner:Well, on the flip side too, you know, vibe coding my public trading bot was really fun, and I learned a lot about trading, to be honest. So I think that is a cool way to build new features and some of the things you guys mentioned and how AI is gonna have an impact on all the data flow of of financial information.
Emily Kurtz:Totally. And obviously not the topic we're talking about, but it's all it's also related to the hot topic of NCP right now.
Jordan Metzner:Yeah. That's for sure. Yeah. Being able to connect to the datasets and and being able to make
Emily Kurtz:decisions that way. That and communicate with LLMs in, you know, a clear way.
Jordan Metzner:Great episode. Just wanna say thanks to Emily and Jake and the entire team at Public for helping us put this together and getting us access to the public APIs so that we could build our prototype this week. It was super fun and learned a lot about stock trading. Don't forget to subscribe and like and click the notification bell. And thanks, everyone.
Jordan Metzner:Great episode.
Sam Nadler:Thanks, everyone.
Jake Trefethen:Awesome. Yeah. Thank you, guys.
Jordan Metzner:Bye, guys. Thanks. Built this week,