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.
Okay. So pretty rhythmic, kinda cool idea. And then, you know, as the store gets busier, and you can hear, you can see the occupancy. And then, as you keep going, the party gets higher and higher. And then as we get to the top, we're gonna a closing mode as well.
Jordan:We'll show kinda how we close it out, and they'll bring music back down. Chill. So
Sam: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 Nabler, cofounder here at Rise Labs. And each week, I'm joined by my friend, business partner, cohost, Jordan Metzner. And this week, we have a special guest, Angie, the CEO of Standard AI. Welcome, Angie.
Angie:Thank you. Thank you. Happy to be here.
Sam:Tell me just in one minute or one sentence a little bit about yourself and a little bit of Standard AI, and then I'll get into kind of our overall docket for the week.
Angie:Awesome. Yeah. So I'm Angie Westbrock. I'm the CEO of Standard AI, and we are using next generation cutting edge technology and computer vision to bring the type of metrics to brick and mortar that were once only available online. Super exciting.
Angie:I I love building, and we have a great team. So this is a awesome place to be right now with, you know, a lot of emphasis on generative AI. We like to talk about computer vision and all of the awesome things that can bring to brick and mortar retail.
Sam:Great. So this week, as always, we're gonna cover a tool we built in the last week. However, this week, we built it with Standard AI in mind, and we think you'll yeah. Maybe it's a potential product feature that Standard AI can adopt, but it's intended to be fun and just hear your feedback. Then we're gonna get into really what Standard AI does.
Sam:And lastly, we're gonna cover the latest greatest in AI news. Gamma had a massive fundraise. We've covered Gamma previously. SoftBank, has some interesting pivots, from NVIDIA to OpenAI and, a little bit of Eleven Labs, one of our favorite tools. So, before we jump in, Jordan, any thoughts on on the day?
Jordan:Yeah. Super excited chatting with Angie. Thanks for joining us. Obviously, another exciting week in AI. Not a dull moment and lots of new announcements from all top tier companies.
Jordan:So, yeah, it's been an exciting week. And, yeah, let's jump right into it.
Sam:Alright, Jordan. I know you're excited about this one. You know, we've done a few music inspired tools. So why don't you walk us through what you built?
Jordan:Yeah. Well, I think, you know, the idea was originally yours, so I just tried to take your idea and kinda put pen to paper a little bit. But we try to keep of think of standard AI in mind. So I've been loving generating music with AI. I love to use Suno.
Jordan:I've tried Oudio and eleven Labs APIs and some other tools as well. And I think this is a really cool way to democratize music production because, you know, now a kid in in their bedroom just like they could with GarageBand, but now, even younger can just describe the kind of song they want and can make whatever they want. And I think that's super cool. And as we know, standard AI really just focuses on retail. And so we thought, you know, how do we bring music and computer vision and AI altogether to build a really cool experience that could be something standard AI would do.
Jordan:So I don't know if you ever used the DJ from Spotify, but it's, you know, it's an AI DJ. It takes kind of like, you know, the best of playlists and and kinda plays them back to you, and then he speaks to you every once in a while, which is sometimes cool and sometimes annoying. But, you know, we kinda took that idea and took it to the next level. And so we thought, you know, I thought of a a retail store, like a clothing shop. I thought, you know, the music should be DJ'd and kind of varied throughout the store.
Jordan:But, you know, most clothing stores, you know, they usually pay to get like popular music and they just play whatever's on the radio kind of thing type of music. It doesn't really matter if the store is busy or not busy. It doesn't really matter if it's morning or night. It doesn't really change if it's cold outside and raining or if it's, you know, a sunny day. They just kinda, you know, play whatever's popular or whatnot.
Jordan:So we thought we could use AI and computer vision to kinda build a DJ that would play different music throughout the day depending on the occupancy of the store. So we built five different songs. You can see the songs that we built down here. So they go from 90 BPM all the way up to one thirty and back to a 100. And all of these songs kinda map to how many shoppers are in the store.
Jordan:So we've got kinda from opening mode all the way up to like full packed store, and then all the way down to closing mode as well. And so let me just jump into this demo really quick. Any anything I missed, Sam, before I do that?
Sam:No. And our our thesis is that, you know, there could be a correlation between beats per minute and, I guess, sales? Or, you know, what would be interesting is does in, music influence, sales, and could a retail location, you know, try and drive sales by just changing music. And it could be the the total opposite. Maybe more chill music, or it could be based on what they're selling.
Sam:You know, a candle shop may not want a 130 beats per minute, but sunglasses may. So either way, we we think it's a a fun way to to think about retail volume and correlating that with with music.
Jordan:Okay. Cool. So let's get into it. So let's start the music. And even really cool here is the intro song has even like an a welcome message.
Jordan:Maybe we can get to that. Let's see if we can get there. So she like welcome shoppers or something like that. Alright. So here's the 11AM beat.
Jordan:And then as you keep going, the party gets higher and higher. And then as we get to the top. Closing mode as well. We'll show kinda how we close it out, and they'll bring the music back down. Chill, so.
Jordan:Cool? Alright. Alright. So, yeah, Angie, love to get your first thoughts on our retail mood AI DJ mixer.
Angie:I love it. I'm a sucker for music though, so I'm an easy sell. But I think something that you said, Sam, is really interesting, which is you have a hypothesis about what's gonna influence shopping behaviors, and that's exactly where this would combine with standard AI technology because there's lots of things that impact shoppers. Customer experience is super important and even more relevant when you think about how to differentiate what's happening inside a brick and mortar store. So I love the idea of being able to play different music, see the response on shoppers.
Angie:And, you know, our what we do is you can't always see an instantaneous reaction on sales data alone. So we track what shoppers are actually doing and how they're responding and if that creates more browsing behaviors or not. So I kinda love this idea, and I think it'd be awesome to put it in one of our stores to have the technologies so that we could we could see whether your your upbeat music and your hypothesis is right.
Jordan:Awesome. Yeah. We even thought to take it to, like, the next level where, you know, essentially, you can see which tracks within, you know, that substrate do perform better or worse. So, you know, hey, disco works better in the afternoon, and we see people shop better, but this disco song just performs marginally less than this other two disco songs. And you can kinda keep creating iterations to kind of create improvements, you know, until you get a disco song that kind of like, you know, really maximizes your yield.
Jordan:And maybe that doesn't work in the same kind of different types of vertical, and then, you know, just the same across all different types of music spectrum. So, you know, feeling out the vibe of the store and
Angie:You could have personalization there to some extent, you know, where you Absolutely. Could know somebody in a certain aisle behaving in different way. Maybe you could even tune in to preferences, like, if it trained over time. I think it's awesome.
Jordan:And just the idea to have music that's unique to the store and the brand. I mean, you know, it's like you go to you you go during the holidays to the shopping mall, and all the stores are playing kind of the same music, you know, over and over again, kinda and I think that's, like, the demonstration that, like, there's no data. Right? Everybody just presses play on the same playlist. And, you know, maybe by generating, you know, Christmas or holiday themed music and knowing which ones work better and worse during different times of the day, you actually could significantly, you know, increase store sales without playing kind of that same Mariah Carey song.
Angie:Thank you.
Sam:Are you telling me no more Michael Buble for on repeat?
Angie:Don't take away Michael Buble. I I just need it at least once.
Sam:Well, I think that gives us the perfect segue into the overall Standard AI product. That was just for fun. But you mentioned something that really resonated with me, which is kinda referring to Standard AI as, like, Google Analytics for physical stores, and that almost immediately cemented the concept for me. But, you know, for our viewers, could you just walk us through Standard AI and how it is positioned or becoming the or is the Google Analytics for retail stores?
Angie:So when you think about what Google Analytics unlocked online, which is rapid AB testing, personalization. You know, when we started especially when I joined Standard, I mean, you think about physical retail and brick and mortar, it's just been this black box. Right? So you just you don't have access to what's actually happening inside the store. You're only able to look at sales data, which is influenced by so many things, you know, seasonality, traffic, weather, you know, all these different patterns.
Angie:So if you're in the business of, you know, store layout, merchandising, marketing, media, things that you're leveraging inside the store, very, very difficult to know how they're performing with real data. And here now, because of computer vision and being able to deploy right on the security cameras, we can unlock that same level of dataset, that same data dataset, that same level of fidelity that you would be able to use to drive online decisions and do it inside the store. Store. So really exciting, especially for retailers that have both an online and in store presence. Now they can start looking at metrics that are similar in both places, and it really sets the stage for, some cool experience changes that I think will come in the future around personalization and just where it goes next.
Jordan:That's super cool, Angie. Just, just a few follow-up quick questions. How does this impact kind of speed? Kinda how fast are retailers able to get access to this data compared to, you know, how long it would take them to get access to something like checkout data or other types of data to make actionable behavior?
Angie:I'm like a operations nerd. Okay. So the thing that I preach over and over again to teams over the years is the importance of leading indicators and metrics over lagging. And so when you think about the delay, it's not just the delay in getting the the transaction log data, which can be weeks down the road. It's also just what is it's it's such a lagging indicator of what's happening that you can't get that crisp crystallized feedback on how the shopper is responding.
Angie:And so not only can we deliver it faster, but you can get to statistical significance way faster than you could by trying to use sales data. And a great example of that is if you talk to almost any retailer, if they're trying to set up an AB test, it's very, very difficult because, you know, to set have a controlled environment. I you know, perhaps there's nothing more dynamic than physical retail. It's also why I love it, but very, very dynamic. So oftentimes, retailers will set aside a store set of, like, 40 stores, let's say.
Angie:And they'll try to make 20 a control, 20 of the experiment. They have to run that often for something like twelve to sixteen weeks to get enough data to be able to analyze it. And even then, you know, you talk to any any anybody in this space, and they'll tell you, you kinda have to squint and really study hard to see if you can correlate a lift or any happen directly with your decision. And you think you know, with ours, we've had we've had some of our retailers and our customers be able to get to statistical significance on feedback of a change within three weeks and using only a couple of stores in some instances because they have so much confidence in what that change is doing, seeing the reaction that shoppers have versus only relying on sales data that they can then scale those insights across the board.
Jordan:That's awesome. So they can experiment faster and then deploy the successful results of the experiment faster, thus, like, yielding higher
Angie:results? Exactly. And then you think about the importance of that time factor, you know, in the case where the change that you made was successful and it made the desired effect, waiting sixteen weeks to implement it is, you know, a bummer. Right? But the bigger problem is is what if it created an adverse effect or had a negative impact?
Angie:Now you've waited sixteen weeks to see that, and even in your stores that you're testing in, it was a negative experience for those customers during that time. So, you know, we think that providing this level of, like, rapid feedback in the physical retail environment is total totally new unlock that they haven't been able to do before.
Sam:Super cool. I have I have two questions, and this may be pretty basic. But how does it actually work? Like Yes, you know, what what what are the retail locations purchasing? What what's the setup?
Sam:And then I we did discuss previously this really cool metric called the visual engagement score. I would love for you to kinda walk us through what that metric means and and how it's driving change for for these locations.
Angie:Yeah. So first, let me just make sure that I say it's all it's all privacy built, privacy first. So we don't use facial recognition. There are some vendors out there that that incorporate that as part of their tech stack. It's always been a pillar of ours to be completely private.
Angie:And and so what we do and what our our proprietary algorithms do is we label it's, like, 26 parts on someone's body, and we put together if you see any of our marketing materials, sometimes you have these little stick figures walking around. And that's because that's that's effectively what we see. We turn all of the customer or any human in that space into effectively, you know, a a digital stick figure. And what that allows us to do is when you compare those those places that we're we're marking, you can distinguish people from one another, but you can't identify who they are. So those you know, those parts that we're labeling, they're not like a fingerprint.
Angie:You can't identify it out of context. And and so we can do lots of really cool things with that because instead of just knowing that a human is in frame, we're actually following how they're moving, where they're where they're gazing, how they're looking. And so the level of confidence that you can have in that data is just a lot more powerful than, you know, just there are some vendors out there or suppliers that will just look at dwell time, for example. And they'll they'll infer that if you stood in front of something, that you must have been engaging with it. For us and Sam, that visual engagement score is really a way to, in a physical store, bring what is like a digital impression on your site, and you think about all of those engagement engagement metrics that you think about.
Angie:Did someone view your content? Did they click on it? Did they share it? Right? There's, like, a level of engagement that's different.
Angie:That's what we bring. So this is a really interesting and important distinction if you're trying to determine how that display is performing or that new product. Like, the the new product example for me is a really interesting and one, because it captures exactly what we're talking about here. You introduce a new product into market. If it's selling, then great.
Angie:Right? But I think it's, like, more than 85, 90% of the time a new product fails, and all you do is get the sales data that says it didn't sell. Well, did it did people see it and not like it, or did they just not even have awareness of it? So we can give you that that, you know, mid middle of the funnel metric to tell you where to invest to make it work. So if you're getting a lot of engagement and it's not converting, that's telling you you've got a product problem or a pricing problem.
Angie:Right? But if there's just no engagement, no awareness, now you know it's placement, marketing, you know, those signage, those kind of things where that's in the right place, you know what you're doing, and you can see those results then faster. And you put that in you know, that's just one example of many where you can apply that type of data to just make better decisions faster so you're investing and and ultimately growing revenue.
Jordan:Obviously, as you mentioned, kind of the company started as, like, just walk out and now kind of focused on kind of bringing analytics to retail a little bit. But, you know, how has AI had a impact over that time frame since you guys were kind of early into the AI AI business, obviously doing a lot of computer vision, and kind of, you know, how does AI have an impact today, and kinda where where do you think it's going on the impact of your business?
Angie:I mean, the the breakthroughs in computer vision are what enable this technology to even exist, you know. Security camera footage has been around for a long time, but the problem is you just could never process that much data in a way that made it consumable and usable. And so without AI and without these breakthroughs in how we are, you know, able to process data and and and use it, none of this would be possible. So it's it's really incredible to see what that facilitates and allows moving forward. And, you know, we didn't even get into the the sort of predictive piece of it, but, like, the longer that we're running in a store because it's you know, because we have AI, we can start to run predictive models and make more predictive, you know, behaviors in terms of what is gonna drive those sales or or simulate making changes instead of having to even do them in a physical physical store.
Angie:So the not only what it unlocks now, but what it's gonna do as we continue to iterate and change and and bring new technology into physical stores is just really incredible to think about all we can do.
Jordan:That's awesome. Really cool product. Super exciting. Seems like just the early days. And, yeah, probably it'll lead to better customer experience in the long run of more personalized, more tuned shopping to to what customers love.
Jordan:Right?
Angie:Yeah. Let's get them let's get every you know, the last thing I'll say on that is too, I mean, I think there is a little bit of a misconception about in store shopping. You know, most retailers that you talk to are still experiencing something like 85 to 90% of their sales are still happening inside the store. So this is an area that we need to be investing in. You know?
Angie:And as a consumer, I I love going in and being able to interact in the store and and make those decisions live, and I I really appreciate the merchandising that happens inside stores. I'm definitely a sucker for impulse buys. So, you know, anytime that we can continue to invest in that the retail industry, which is incredibly important for lots of reasons for the economy, I think it's it's a great place
Sam:Amazing. Amazing. Okay. I'm gonna transition to three news stories. The first one, I think, is exciting.
Sam:One, because we've covered Gamma as one of our favorite product tools. But Gamma, which is a a really cool AI tool that allows you to build beautiful presentations within seconds, you know, a really simple prompt. It'll kinda build the outline and then design the presentation, almost completely for you. There is usually a little bit of a a prompt tweaking after that, but just raised a massive series b at 68,000,000. And, apparently, they're profitable at a $2,100,000,000 valuation with a really lean team of 50 people.
Sam:Jordan, I know you've used Gamma. Angie, have you ever used Gamma?
Angie:No. I haven't used Gamma, but I've used some of the similar products in market too.
Sam:Cool. It's fun. I recommend checking it out. But, Jordan, any top of mind thoughts as I know you're a you're a fan?
Jordan:Okay. Yeah. Let's first talk about some of the stats. So $2,100,000,000 valuation led by Andres and Horowitz, 100,000,000 in ARR, a team of just 50 people. Let's say, according to them, 2,000,000 in ARR per employee, which is incredibly large.
Jordan:You know, this is a market some of the other numbers are crazy too here. 70,000,000 users and 30,000,000 Gammas created every month. I mean, these are just bonkers numbers. Okay? First of all, PowerPoint was is a flat flat presentation format.
Jordan:You would expect AI to just kill PowerPoint. This has been completely dominated by Microsoft historically until Google Slides came into the picture. You would expect Google and Gemini and Microsoft with their copilot to present like something similar to the market to be competitive to this. And I think it just shows like how fast small startups can build a niche inside a market like, you know, presentations and slide decks to build just a massive user base. And yeah.
Jordan:I mean, it seems like Google Slides and Microsoft will probably still go after Gamma, but 70,000,000 users is just I mean, that's an incredible amount of users for a early stage company, and and hats off to them.
Sam:Yeah. I think as, you know, someone who's in the start up building space, there's always the question that we get whenever we're launching a new product that's can't an incumbent build that? Can't Google build that? Can't Microsoft build that? And, yeah, they they probably can, but I think Gamma's an example of they sometimes won't or, are, you know, thinly spread or, don't have you know, can't innovate fast enough.
Sam:So I think Gamma is a great example of, how you can disrupt some of these major players where it would theoretically seem easy for them to build.
Angie:Yeah. I think these are I'm I'm so fascinated with there's, like, a few of these companies, especially, that came out right in that 2022, 2023, you know, generative AI launching pad that happened then that recognized if you if you leverage AI for because I really consider this not just like a tool to disrupt PowerPoint, but it's it's like a workflow automation and content creation. It's it's so much more than that. Right? It's like completely reimagining how you build a slide deck.
Angie:And I think that's why, you know, we're seeing that those types of, you know, innovation in those areas are are getting adopted. It's just it's such a much better experience than what we had before, and they know they're a customer. They're gonna be completely focused on this product, which I think is what makes them be able to stay ahead of the the big the big ones. Right? And, you know, the other thing I'll say that I loved about, once again, the operator, Nerg and Me, but, like, you know, they talk about profitability.
Angie:That is still relatively new in the VC row. I mean, it was only a few years ago when it was growth at any cost, and now you see these, like, really lean teams accomplishing awesome, awesome innovation, and doing it with a focus on, like, the business fundamentals at the same time. Awesome.
Sam:Okay. Cool. So next up, SoftBank sells their entire NVIDIA stake and pivots to OpenAI. What I found interesting is this obviously, it seems a bet towards the application layer. I've also heard that this isn't the first time SoftBank has sold their entire NVIDIA stake.
Sam:It's their second time in recent years. And I did hear if they had maintained their NVIDIA stake, not selling the first time, it would be worth over 200,000,000,000, today. But either way, I think the news is interesting, the pivot away from from the chips and towards the application layer.
Angie:Yeah. I mean, like, this is sort of classic top bank. Right? I I they they like to be on the the forefront of what's coming next, you know, and and they're they're making a bet on application layer. But, look, I mean, I still think that Nvidia is a good bet too.
Angie:Know? Yeah.
Jordan:I mean, we've seen SoftBank gamble a ton. Obviously, they're a huge investor and owner in AMD. And he, you know, he's made a ton of gambles in venture capital and has had some big wins, some big losses. You know, I think one thing it does say, you know, ignoring that, you know, he's moving cash out of Nvidia, I think it it does kind of put even more pressure and spotlight on OpenAI. And it seems like just over the past kind of, you know, I would say, like, ninety days or so, it seems like OpenAI is now being considered a hyperscaler in the same talks of, you know, a Google, a Facebook, and Amazon.
Jordan:And, know, I don't know how far away they are from going public, but it it certainly seems like they're, you know, kind of joining this, like, mag seven crew. It doesn't seem like anyone else is nearby, and I think that, like, you know, this might be the final kind of solidification of that prior to, you know, going public. Alright. Let's last one, and this one's fun, and then we'll move on.
Sam:Okay. Last one is Eleven Labs launches the iconic voice marketplace. Jordan, I know you're a huge fan of Eleven Labs. And my understanding of the iconic voice marketplace is it allows brands to license AI generated voices. So maybe the Matthew McConaughey voice or, who else has an iconic voice?
Sam:Maybe Muhammad Ali. I don't know. I'm just trying to think through. Maybe that's not on the list, but, kind of a cool product and and a fun one to talk about.
Jordan:Yeah. I think everybody's looking for, like, Samuel L. Jackson or Morgan Freeman, who are both not on the list. But some of the fun voices are Michael Caine, Burt Reynolds, Maya Angelou, Amelia Earhart, Babe Ruth, Mark Twain, Ty Cobb. I think those are kind of fun.
Jordan:Thomas Edison, Mickey Rooney. You know, I think, look, it's just the early days of, you know, basically talent being able to leverage their skill set, you know, with AI to monetize. And, you know, if you think of some of these people on this list, a lot of these people are dead. And that means, you know, they were able to get licensing from the estate, and this is like a new economic, you know, economic revenue stream for the estate of, you know, many of these folks. And for those who are alive, obviously, they probably agreed to this.
Jordan:But, yeah, I think it's a cool way of kinda combining talent with AI. You know, we we've seen so many weeks in a row, Sam, we talk about kind of how, you know, Veo and Sora are kinda killing the actors or killing, you know, Hollywood or all these things. This is kind of a a rejuvenation of kind of them being able to leverage their persona with AI to kind of remonetize again.
Angie:Yeah. I also think it's reinstatement and and, like, where consent and verification is going. Right? Like, I loved reading about this because I it's sort of an area I hadn't really thought much about before. You know?
Angie:Like you said, like, how to make these marketplaces and monetize it. But then the operator immediately goes to once you are giving your consent on using your likeness, likeness? Like, how does that how does that stay constrained and, like, you know, how are they gonna manage that? It's probably why we see more people that are historical figures than still alive because there's, you know, there being some hesitation. But inevitably, this is where it's going.
Angie:Right? And and I think it's a a really interesting way to to bring consent at the forefront of AI creation.
Jordan:Yeah. It's just super cool. I think bringing it back around again, you know, whatever you can get, you know, fat man scoop to be a DJ, and then you could put him inside the stores, and then, you know, you have fat man scoop, you know, DJ ing it up and so, you know, maybe you can get Diplo and you can get all the top DJs and put them in all your retail stores and there you go. You got kind of an AI AI DJ for every occasion, Angie.
Sam:Awesome, guys. Great episode. Thank you for joining us, Angie. And to the both of you, any final thoughts before we wrap up?
Jordan:No. Just wanna say like and subscribe. Check us out on YouTube. We wanna say thanks to Angie and the Standard AI team for their awesome technology. Check them all out online at standard dot a I.
Angie:Thank you so much. This was so fun. I love the format of it, and we gotta figure out what to do with that. Yeah. I generated music.
Angie:I don't know. I think you're onto something there.
Jordan:Awesome. Thanks, Angie. Absolutely. Thank you so much.
Sam:Thanks, Angie.
Jordan:Bye, guys. See you next week.
Sam:Next week, joining us on Built This Week is axiommath.ai.