We Built an AI Tool That Replaced a Week of Work

Sam Nadler: Typical turnaround time to preparing, you know, a seven or eight different estimates for for an off-site would take us about a week and a half. I would say now it takes us three to four days.Jordan Metzner: Built This Week, breaking it down. Built This Week, we show you how. A fresh idea, a clever tweak you locked in. You built this week.Sam Nadler: Hey, everyone, and welcome to Built This Week, the podcast where we share what we're building, how we're building it, and what it means for the world of AI and startups. I'm Sam Nadler here, cofounder here at Rise Labs. And each and every week, I'm joined by my friend, cohost, business partner, Jordan Metzner. It going today, Jordan?Jordan Metzner: Hey, Sam. How's it going? Big week in AI. Like, every week, obviously, we saw some major models come out from some hyperscalers. And, yeah, lots to talk about this week.Never a dull moment in the AI world. So, yeah, looking forward to this week's episode.Sam Nadler: Exciting week. And before I jump into the docket, don't forget to like and subscribe. We have new episodes every Friday. I think we're just at or about to surpass 25,000 subscribers on YouTube. So what a milestone, Jordan.But, yeah, hit that like and subscribe button and and get notified. This week, Jordan, we're gonna jump into a tool we built for one of our portfolio companies that so actually a pretty simple web app, but it helps us keep organized and plan events a lot faster than we really historically have. We're also gonna dive into a tool both you and I love to use for prototyping. It's just the Google AI Studio Suite. And lastly, we're gonna cover some news from Meta and Anthropic.Anything top of mind before I I demo our little tool?Jordan Metzner: No. Let's jump into it. Think we got a lot to show today, so let's go.Sam Nadler: Cool. So just as I get the screens ready, just a little context. So one of our portfolio companies built, you know, Pre AI is a corporate travel company. We help plan off sites and other type of corporate events for all different types of startups. You know, we do them from anywhere from, I would say, 20 people to 200 people, if not more.And there's a lot of little moves moving pieces that go into planning a large off-site or a conference, whether it's the food and bev minimum, whether it is, you know, AV setup, conference room fees, making sure every everyone has an individual room, etcetera, etcetera, etcetera. And one of the kind of value propositions of Offsidyo, the portfolio company that we're talking about, is we once we quote quote a price, that price never changes. So it is really important for us to have every single cost element identified and and planned prior to quoting a price. Otherwise, we could find ourselves underwater where, you know, maybe we quoted, just to make the math easy, a $100,000 for a 100 person off-site in in, you know, Austin, Texas. You know, if if that off-site ends up costing us a 110,000 or a 105,000 or a 120,000, we're upside down on that event, which would be a a big problem.So prior to AI, and, you know, we we did a lot of this in evolving spreadsheets, and our spreadsheets became even more and more sophisticated. But, you know, with the invention of AI, I built a little tool that allows us to do this very, very quickly. And I call this the again, I'm not that great with names, the Offsightio estimator. So just to walk you through it, this is a this is a demo one. This is not real data.It's not a real client. But, you know, basically, can choose a new estimate, and you go in and filling out all the information for that potential estimate. One really cool and I'm not gonna do this. I'm gonna go back to a current estimate we already have completed just to because it shows exactly what you know, we could go into this Hilton grande estimate in Buenos Aires. And so it pulls out one is you go in and just, like, add all the necessary information you get from the potential bids.This could include meals. This could include different activities. This could include meeting space, AV, you know, and it just really helps you in a very structured and organized way and make sure you have all your pieces lined up. But the really important thing is that it flags potential pitfalls like food and bev minimums. You know, there's certain there's a lot of little things hotels, if you're not totally used to negotiating with them, that they can they can kind of sink their cause into.One of them is the food and bev minimum. Other is AV cost, etcetera, etcetera. So we built this tool in a way that will flag, you know, if certain costs haven't been considered. So we lock them in with the hotel prior to communicating the cost with the client. So this is you know essentially replaces a very elaborate Google spreadsheet that we've you know, used over the years to to to perform this exercise in a very structured and easy to digest way.However, what's really fun about this is, you know, post confirming different locations and their prices, we would then need to present these different options to our clients. And that would take a lot of time in in corralling all the information and then, you know, creating either a Canva or a Figma presentation to, you know, display it in the best way possible. Each one of these presentations would and and to some degree, take us days to to perfect. Now we created a different tool that will just pull the information from these estimates that we created of the different properties. You can upload the client logo.You can you can choose which estimates which properties to include in the estimate. Maybe you did 10 estimates, but you only wanna include seven properties. So you can just, you know, choose which one, and then you can essentially go create immediately the presentation, and you get a website that is you can share with the client instantaneously. So this is obviously demo data, but, you know, immediately, you would have everything needed for this potential in off-site location. The client can see the map, what's included, the all in per person rate, the dates, how many nights, how many guests, etcetera.And they can even you know, once this link is shared, they can shortlist which ones they're they're most interested in. So theoretically, if you present them seven, eight properties and they're like, hey. I'm most interested in these these two in Miami and this one in New York City, whatever, they can shortlist them. And then on in our next call, we can engage specifically on those three properties. Overall, it's not that crazy of a tool.However, I will say it has saved us a lot of time. I think it's probably our typical prior to having built this tool, our typical turnaround time to preparing, you know, a seven or eight different estimates for for an off-site would take us about a week and a half. I would say now it takes us three to four days. As soon as that information comes back from the hotel and we've locked in those rates, we can essentially have a presentation ready for our client. So it just greatly accelerated that turnaround time.You know, when you have a client who's potentially interested in off-site or really interested in purchasing anything, speed matters. So that's the tool. Any thoughts? Any high level? And this is live in production.So any high level thoughts or feedback?Jordan Metzner: No. I mean, awesome. Tell us a little bit about how you built it.Sam Nadler: Yeah. Cloud code, I think, initially. And then after some different neat bugs and updates, we I transitioned to Codex, Supabase for the back end. And, yeah, that's pretty much it.Jordan Metzner: And front end?Sam Nadler: Front end. I mean, I think it's just a React web app with Cloud Code.Jordan Metzner: On Amplify?Sam Nadler: Oh, yeah. Yeah. On Amplify.Jordan Metzner: Cool. Super cool.Sam Nadler: Yeah.Jordan Metzner: And and has customers said anything about it so far?Sam Nadler: Yeah. I mean, the good news is is we demoed this for a repeat customer who's been using us for years. And I think the transition, you know, to kind of a well designed website that they could click through versus a PDF presentation that we built in Canva was an upgrade. And this, you know, I think, obviously, this customer's their their own business is growing, but they've booked their biggest off-site with us yet happening in June just a couple months away. So that was you know, that happened at the beginning of this year was the first client we demoed this this technology with.And, you know, now it's kind of just every we we're using it every day.Jordan Metzner: That's amazing. That's incredible. Alright. Well, great job. Honestly, it looks great too.What do you any future plans for it or any other, like, AI features you wanna build?Sam Nadler: Yeah. So, you know, we we're working on kind of v two and v three. Maybe it's all part of v two. You know, we use cVent to generate bids. We would like to have that all incorporated in this tool so we could automate the bid process as much as possible.We also have certain capabilities for the client to log in and see the progress, but I think there's a lot of opportunities for us to further build that out where there there's this instead of just being a internal tool for us and a presentation tool, it's also a tool for the client. There's lots of communication and choices that a client will make post confirmation that they're going on the off-site. You know, whether that's the rooming list, whether that's dietary choices, whether that's exact itinerary choices. You know, a lot of that isn't fully built or decided on until actually days before the off-site. So giving them a central location to go in and confirm and see where they're at and what they've chosen so far, if any adjustments need to be made, are things we're working on now.Jordan Metzner: Dude, it's awesome. Honestly, great job. Cool. Super cool.Sam Nadler: Cool. Alright. And transitioning a little bit is the Google AI Studio. I know you and I are both, like, huge fans. Do you wanna walk me through anything you've built in Google or how you use Google AI Studio and just, you know, tell me show me a few things you've built.Jordan Metzner: Yeah. Let's jump into it. Yeah. So the website's just a i.studio. I think it resolves to something like a istudio.google.com.You might have to set it up with, an API key so you can pay for your use. It's a little complicated. But once you get it all set up, the reason I like it a lot is that it has a lot of different tools and features kind of all built into the same, like, standardized UI. So, you know, what's cool is over here, can kind of pick your models. So, you know, if I wanted to do something, you know, really quickly, you know, let's just say, like, you know, what is two plus two, you know, I'll get just like a regular chatbot, you know, and it'll it'll respond fine.You know, that's cool, whatever. But if I wanna do deeper research, I can kinda go here and go to, like, Gemini Pro, and then I can, you know you know, have it do some research. I can add some tools. Like, I can upload some stuff or I can choose to turn on Google search. So kinda like ChatGPT esque features, I think, which might be like, okay.That's that's cool. And then but if you go back a little bit and kinda go back to here, what's cool is you can also do, like, image generation with Nana Banana. So if you go over here, you see, like, Nana Banana is now my selected model, you know, and, you know, make me a clay figure, a black lab dog at the park eating a hot dog. And then, you know, it should basically generate that. So now we've got, like, a nana banana in here with all of this, like, you know, preferences that you can kinda come out of with this thing.You can add on Google search. You can add on image search, all these types of things. And again, we're, like, still within, like, the same setting of this, like, AI studio.Sam Nadler: Yeah. Have you used their new music model? I know you and I have both played around a lot with Suno, but I have not used the Google I think it's called Lyria, if I'm not mistaken.Jordan Metzner: Yeah. Lyra. Yeah. So again, this is like another cool thing. So, like, you know, they drop a new model.It's easily available here. This is Lyra three Pro and Lyra clip preview. So clip preview will do thirty second out clips, and then Lyra Pro will do longer than that. I guess, like, the interest of the show, let's just do clip preview. So this is like a really long thing, but, you know, let's just say, like, you know, write a reggae song about the Built This Week Podcast on YouTube.How's that? So that should let us go generate. So this one, again, this is the Lyra Pro preview. So I think this one's the this one's the full full one. Maybe we'll get a full song out of here.And then just to jump back here, we got the dog with the hot dog in his mouth off of the image generation. And, again, I don't know how long this will take, but hopefully just a little bit. But, yeah, just going back to kind of like all these some of these really cool features inside inside the studio. You know, you can also go here and do video, so you can do that. You can do more audio features as well like we talked about.There's also a live feature where you can talk to Google. And then I also has this other cool feature that I actually really like a lot where you can share your screen. So if you need help with certain features or you're trying to build a website or something like that, you know, you can kinda go here, share the screen issue, and get Gemini to help you there as well. Alright. Alright.That's enough. You know, it's a fun little tool, and I think it just shows kind of like the diversity of kind of the Google tools that you have at your diversity at your availability. AndSam Nadler: So how do you think about you know, obviously, you use Cloud Code a lot. You use Codecs. You use Google AI Studio. You know, back in the day, we've demoed things we've built on Bolt and Lovable that, you know, we're not really using anymore. But how do you think about which tool to pull out and which tool to use?I mean, for me, it's like, is it is it kinda throwaway? Is it something that's like a quick and easy, really fast demo I wanna build? I'd I'm typically doing it in Google AI Studio. Otherwise, I'll probably start it in Cloud Code and then eventually transition to Codex or vice versa. How are you thinking about kind of which tool to use?Jordan Metzner: I think it's kind of the same. You know, I think, like, each of these tools kind of, in my opinion, lives in, like, a different vertical. So, you know, if I'm building something that's like a big piece of software, then it's gonna fall into the vertical of, like, you know, cursor, clawed code, or or OpenAI or codex. You know? If I'm building something that I need a quick database and I need it to live for a little bit, but then I'm gonna throw it away, Then it kinda falls into that, like, you know, kinda like Replet lovable ish world.And then if I'm just building a demo to show someone a gimmick, then it kinda falls into that, like, first column of, like, the Google AI studio or lovable or bold or something else to kinda build something quickly like that. But, you know, as you get more comfortable with each of them, you kinda see their, like, strengths and weaknesses. I still really love this Google, like, AI Studio for just building out little web apps because it's it's so fast and so clean and gives you good inspiration. But as you know, all these tools are just building stuff so so well. And even sometimes, like, I use the chat.Like, I use Claude just in the browser to have it build me something as it does a great job there as well.Sam Nadler: Yep. Cool. Alright. Well, let's transition to the news. We got some big news stories.Let me kick off with Anthropic, actually. Anthropic, you know, in the last couple days, I think it was, I guess, released that not released. I guess, yeah, a news release that they have a new AI model that's so powerful that they're not releasing it, that they've, like, generated a a I would say a committee of, you know, dozens, if not, you know, several dozen corporate players to test security initiatives before they release it to the public. But, you know, I think some of the metrics were significantly higher than Opus 4.6. But yeah.I mean, what do you think? A, is it hype? Is it real? Is it a marketing gimmick? And, yeah, any any high level thoughts?Jordan Metzner: Well, I mean, it's funny that this news comes out, like, you know, right after the news of, you know, Anthropic and kind of their relationship with the US government, the defense department. And no. I mean, I I think, you know, I read some further beyond this article about how they created a consortium of 40 companies like AWS and Microsoft and Apple and a bunch of others to come together to kind of solve some of these security issues. And it's not surprising to me that, you know, kind of if you take a very heavy model based on the newest NVIDIA chips and you build something focused on, you know, coding that it's able to find vulnerabilities that, like, maybe the human human being didn't find previously. But my understanding is these are some pretty big vulnerabilities that, like, we didn't know existed.But as these models keep scaling up, you know, they're gonna be better and better at finding them and better better at exploiting them as well. It does make me feel weird that, you know, something this powerful according to them is in the hands of, like, just Anthropic. You know? Like, this is a private tech company. It's owned mostly by, like, one person and his sister.You know, we like Anthropic. We use it a lot. We pay them a lot of money. But if it's as dangerous as they say it is, you know, maybe this is an issue of national security. I don't know.I want my hands on it so we can start coding with it, to be honest with you. ButSam Nadler: I think we'll have to wait a little bit. I think, like didn't they say that I may be mistaken here. It was, like, 2,000,000,000,000 parameters to train this model or or even more, I think I've I've heard. But it's one of the biggest it is the biggest model ever created. What what what was the metric they said?Jordan Metzner: I don't I don't know about the how many params it is. I know that it's, like, I heard it's one of the first models built on Blackburn. And so, you know, we're gonna start to see this. We're gonna see it from the other major major model companies. And, you know, will they all act the same way where if they built a model like this, will they release it in this fashion or will they release it more publicly?Or, you know, is this a generalized risk? I think Google, we saw a few weeks ago, talk about quantum computing risks and speaking with, like, kind of the Bitcoin developers and Ethereum developers to make sure that, like, those weren't able to be hacked by their software or, like, their kind of compute stack. So hopefully, we'll continue to see this, like, good actorship happen across these big companies, but at the same time, it seems like, you know, the amount of frequency this is happening is getting higher and higher as, like, AI is getting more and more powerful on day by day basis.Sam Nadler: Cool. So moving on, Meta debuts its new Spark model, which I think is a pivot from its previous strategy of of an open source model. But did you get a chance to to read this article, and and what do you think of of this news?Jordan Metzner: Yeah. So I thought it was pretty interesting, and I was gonna tell you to go to the website. If you could just go to a new tab, go to meta.ai. Yeah. Just check it out, and you'll see that the, you know, meta's in the game.You know, is it the best model compared to some of the other models? I don't think exactly, but it's pretty fast. It's right now free to you to use, and they're catching up. You know? It says 52 opus and 5.4 or higher than that.So, you know, they know they're in fourth place as it tells you right there. But I thought it's pretty interesting. And, you know, if they start to expose these APIs, like, this could be a really interesting game. And, you know, our competition sparks more, you know, inventiveness and ingenuity, so I'm all about it.Sam Nadler: What's Vibes? Okay. We'll see this here. Oh, it's an image generation. Okay.Jordan Metzner: Yeah. Vibes is image. I think they have video as well. If you click where it says image in the middle, you can make videos. So, you know, this is kind of the Sora ish stuff that they have already built into Messenger and some of the other apps across the Meta platforms.But, yeah, it seems like, you know, slowly but surely, you can't give up on them. That's for sure. So you got it? So just, you know, yeah, one new one more chatbot. And you can download the Meta app.I think it's meta dot a I is the app. They have the chatbot available. It's free to use, I think, for Facebook users. And it's top of the App Store right now, like top five, top 10 apps. And, yeah, the game's on.You know? I mean, it just puts pressure on maybe OpenAI and some of these other guys that are charging $20 a month for the consumer. Maybe, you know, maybe consumer chat GPT per se becomes free with Facebook. So it's still early days. It's, you know, Alex Wang's first Alexander Wang's first model of Facebook.So but pretty exciting stuff.Sam Nadler: Awesome. Yeah. Well, great show, Jordan. Thanks for demoing the Google AI Studio. I hope you found my pretty simple estimate trip estimating tool interesting.Again, I think, you know, as long as we can inspire people to build with AI and maybe, you know, apply AI to their own businesses, then that's a win for us. An interesting news. I would love to get my hands on is it called Mythios? What did they call it?Jordan Metzner: Yeah. Mythos. Don't worry. Don't Mythos. I don't think we're getting our hands on that anytime soon.Sam Nadler: Anytime soon. Alright. Well, anything before we close outJordan Metzner: I think getting your I think getting your hands on it is a mythos.Sam Nadler: Alright. Agreed. Probably better. Alright. Anything else before we close out this what episode is this?Jordan Metzner: This is episode 38.Sam Nadler: Episode thirty eight?Jordan Metzner: Thirty eight. So congrats. Crazy. Yeah. Crazy.Crazy. Great job, Sam. Alright. Well, this is a great episode. Fun to talk about AI.Can't wait to see what's happening next week. Hope everyone can like and subscribe and tell their friends about Built This Week, and see you all soon. Bye bye. Built This Week.