Subscribe
Share
Share
Embed
The Matt Allen Show is your front-row seat to the world of finance, investing, technology, and business. Each week, host Matt Allen breaks down the biggest stories in the markets and sits down with leading investors, entrepreneurs, tech experts, and innovators to uncover insights that matter.
From stock market trends and economic shifts to breakthroughs in technology and the strategies of top business leaders, this show gives you the clarity and perspective you need to stay ahead. Whether you’re an everyday investor, a finance professional, or simply curious about the forces shaping our world, you’ll find ideas here that can change the way you think about money and business.
New episodes drop every Tuesday and Thursday.
Welcome back to the Matt Allen Show. I'm your host, Matt Allen. We have a very special guest today, Ryan Steelberg, who's the CEO of Veritone, which is his company doing awesome things at AI. First of all, Ryan, I wanna say I have a lot of respect for you for the things that you've built in the past, what you're building now, and I'm just super excited that you're here, so thank you.
Ryan Steelberg:Matt, great to be here, and it's great to talk to new investors and old investors, so thanks for the time.
Matt Allen:For sure, absolutely. And one thing when I was doing the research is that I know you're a sports fan and I was a sports fan, and you went to UCLA, right?
Ryan Steelberg:I am a Bruin. You know, we're having a complicated year trying to figure ourselves out, but I am a proud Bruin. What about you? So what
Matt Allen:I I live in Tallahassee, so we actually have y'all's alumni as a defensive coordinator, Tony White. Okay. And then I saw you guys were in the news as well today with all the whole leaving the Rose Bowl.
Ryan Steelberg:You know, that I mean, as a student there, the irony is depending on maybe a beer or two, by the time you actually get to the stadium, you're pretty much exhausted. So, you know, for a lot of people, they don't realize it's like, it's over 25 miles away from the school, the campus to the Rose Bowl, but with LA traffic, two hours. Right? I think, so we're excited. I think, you know, I think there's some people obviously that have that romance connection between UCLA and the Rose Bowl.
Ryan Steelberg:However, I'd much rather be able to be a better draw. I think it's gonna be a better draw for talent coming to the school if we're able to play at SoFi. So at least for me, I'm I'm hoping that our home games will be played at SoFi as compared to the Rose Bowl. Real quick on is that is it just
Matt Allen:is that closer as well? It is. It's closer with a
Ryan Steelberg:lot lot less traffic. So Okay. Significantly closer. I think you'll get a much bigger turnout for students and alumni at SoFi.
Matt Allen:Well, before we go down, me and you could probably talk to this whole sports hole the whole time. So before we get down that rabbit hole, I just kinda I would love to hear about what you did in the past with the companies that you sold to Google and things like that before we get to what you're doing today.
Ryan Steelberg:Yeah. Well, I'm an ad tech guy. You know, as we talked about, I went to UCLA actually to be a kinesiologist. I wanted to, I love sports. I grew up as an athlete.
Ryan Steelberg:You know, I wanted to be a surgeon, work on people's shoulders and knees. Thankfully, you know, a great path. I grew up with computers. As early as I can remember, I'm very thankful for my father who was not just a brilliant entrepreneur, but really impressed upon us about coding and that software was gonna run the world one day, and as a little kid, don't really think about it, but in the mid-90s, the World Wide Web just kinda came on the scene. My brother and I had a great idea to try to serve, basically, advertising around real estate listings, which was one of the think of it as zillow1.0 back in the mid nineties.
Matt Allen:Okay.
Ryan Steelberg:And so we got into ad tech early, built one of the first, you know, larger ad technology companies. We were serving all the ads on like Yahoo and all the glory websites of the past and took that company public in 1999 when I was a much younger lad. So we just, you know, I like to say as you sit around and you become an expert in a field, the technology changes around you, and if you can apply that, and then the same goes for AI, we can talk a lot about that, But and I think you know? So the short of it is, you know, ad tech and growing up in that space was where we cut our teeth becoming experts in all things messy data, and it's really where we started to leverage some of the nascent neural networks for ad targeting and contextual based analysis going back over twenty years now. So ad tech sort of grew into a passion for AI, and then obviously that led up to us selling our most recent business to Google.
Ryan Steelberg:It was called DMARC Broadcasting back in the day, and I was there for a couple years and helped head up a lot of their offline ad initiatives. So I think it's been a journey about data. It's been about a journey about AI, and that ultimately led to the founding of Veritone.
Matt Allen:What what what why did you pivot to Veritone? It well, let's start with this. If you if you had to explain Veritone to a random person on the side of the road who doesn't know much about AI, what would you say it does?
Ryan Steelberg:Veritone is the expert on ingesting or taking in lots of messy audio and video and other forms of data and having machines try to understand what's inside that audio and video. Do you mean by messy? Messy. Well, I'm thinking about sports, we're talking about football, right? If you're watching beautiful four k video of watching the UCLA football game, machines still can't understand what's inside that content.
Ryan Steelberg:So in terms of audio and video, can a machine understand which player's on screen? What words are they saying? You know, is there a logo in the background? You know, if you and I sat back and, you know, humans, and we wanted to describe as much as we possibly can about what's happening in the video at any one moment, think of that as rich metadata. Veritone uses AI to extract that data, that intelligence out of that, okay?
Ryan Steelberg:Now that's clean data, but we also do a lot of our work in the public sector. So we work with police agencies and sheriff departments where a lot of their camera feeds are dark and dirty, right? If they're not pristine high quality four ks, but their security cameras, you're trying to identify, you know, maybe a suspect car in a dark alley. So the AI has got to be good enough to understand what's inside the audio and video if it's pristine and clean four ks, but also can understand, you know, poor lighting and fuzzy, you know, bad audio quality, etcetera. So that's what we do.
Ryan Steelberg:We work with hundreds of customers who have audio and video assets or data problems, and we use AI to fix those.
Matt Allen:So with, I know I know one of your clients is ESPN. So what is how how do they do they monetize this unstructured data, or is it just does it make it them more efficient? Is it?
Ryan Steelberg:Yeah, great question. For them and a few others, it's multifaceted. So for ESPN, we actually ingest, we take into our hosted solution for them every piece of content they produce, their podcasts, their linear video, sports center, the actual video show you watch every night, all their programming. We ingest it. We use AI, hundreds of trained and tuned AI models to again, get understanding, right?
Ryan Steelberg:Tagging, everything that's going on in that audio and video. And the purpose of that, number one, you're right, is efficiency gain. They no longer have to have an army of people literally manually watching and tagging, like creating like log files of what's happening. Our technology, our machines can create that intelligence. And then once you've created that metadata intelligence layer, then it really can advance other areas of their optimization.
Ryan Steelberg:So for example, if we're watching a sporting event, you will often hear the host talk about GEICO in a commercial, right? With his voice, like the play by play, and then you'll see like a GEICO logo in the background. What's that worth? How do you evaluate the value of that sponsorship? When you use our software to index it, right, and we get an identify, yes, that's how long that logo is on screen.
Ryan Steelberg:That famous host, Tom Brady, right, when he's talking about Playboy and he mentions GEICO, we can now help create data to evaluate how valuable that ad space, that sponsorship is. Other big use cases is, you know, frankly, programming and ratings. Are certain hosts working or not? Like when they, you know, some of these groups, they have to evaluate. Like they're investing millions of dollars into this talent.
Ryan Steelberg:How is that impacting ratings? So if you were using our software and I can tell you how long Tom Brady is on screen, right, are you seeing an increase and decrease in tune in or tune out? Are you, you know, when that one, you know, and so just think of what Google did to the open web. When create an index so you can search and learn about everything on the web, think about all the things you can do in different use cases when you indexed audio and video at scale. So I just gave you a couple is, yes, they both save money, helps them be more efficient to run their organization, but it ultimately helps them generate more revenue, fine tune their advertising opportunities for their advertising clients, etcetera.
Matt Allen:That's incredibly interesting. And so, well, kind of just a random thought here is that I'm in the, we have a financial media company and we distribute a lot of content, mainly all short form, either through short form, reels or, long form, sometimes on YouTube content, things like that, just with our different creators that we have. For example, let's say you wanted to do a, Veritone, wanna do a retail investor campaign. Well, we would, we would get, we would get our creators, either in house or part of our network and create content through that. I'm a firm believer that even though I watch Squat Box on CNBC every morning, that eventually the creators will be the number one forum media.
Matt Allen:Is Veritone kind of moving toward creators and things like that as well?
Ryan Steelberg:A 100%. You know, when you when you think about, you know, indexing content, I I think, you know, let me say it this way. I mean, the majority of the content that's really moving the needle has you know, it's either a host or an influencer or a creator. Right? Podcasters, you know, and they're and they're they're multi platform.
Ryan Steelberg:Right? You know, this The format you and I are doing today, obviously, will work for a podcast. It'll work for redistributing as short form or long form. How can we find out and investigate and interrogate this conversation? Are we finding out, Matt, that this conversation is working because I'm talking about real use cases for AI?
Ryan Steelberg:Right? A machine can understand this conversation and propagate that out fast. So, again, I do think that AI indexing and providing a better voice for influencers can arguably compete effectively against the CNBCs and the squawk box of the world. Because again, I may have my interest, but I don't have necessarily time to search and discover, right, and find Matt out, but if AI can index the topic that you and I are talking about, then that content can be brought to me in my feed, right? So, I think it's a very exciting time for the creator influencer economy, and I think this ranges from how investors wanna get their information, right?
Ryan Steelberg:How do you really understand what kind of company you should be investing in? What makes them new and novel? You're gonna wait for just global research analysts, or are you gonna listen to certain influencers who may do nothing twenty four hours a day, but just have a passion, right, for advanced AI or AI and robotics, etcetera.
Matt Allen:Yeah, for sure, and I've noticed that kinda like in my life is that I came from, I put out a lot of research now, so I have my own creator. Even though we have Bean Media, which is the company that does what I just said, I still act as an influencer for it, where I put out research and things like that. And AI has been incredible for helping me research companies a lot faster. It hasn't necessarily replaced me, but it's been a tool that's unbelievable. And Veritone seems like it's on the cutting edge, definitely on this side of media for doing that as well.
Ryan Steelberg:Yeah, I think it's, you know, again, we cut our teeth in media and entertainment, and thankfully, you know, we grew up a large enough business and working with the largest media entertainment brands out there, again, from the CBSs and the Disney, ESPN's, etcetera. That same technology that's analyzing the sports content we were talking about is now empowering the US government and state and local law enforcement. Right? And we're helping find bad guys faster, finding that car that may have showed up with a dent on the right back fender. Right?
Ryan Steelberg:And so, again, radical different use cases, but the same underlying technology. And so we're really excited that we can, again, be It's very cost effective and it's frankly why our operating margins are so competitive, right, against other companies out there in the spaces. We can do this at scale. We have been doing it for over ten years, so we can do very profitably, and we've shown efficacy that we can apply this use cases that generate real ROI fast. This is not POC or novelty.
Ryan Steelberg:I like to say is almost every one of our customers, Veritone is now a critical software component of their day to day operations.
Matt Allen:What would you say is the number one ROI that they get? We talked about it little bit, but if you're a creator like me, what would you say it would be?
Ryan Steelberg:Well, for you, it's probably efficiency. Again, I mean, I think we provide probably the most value. I'll kinda hijack the question a little bit and spin it to me for a little bit, but I think we provide the most value for companies and entertainment shops in the media space that are producing or have to deal with thousands and thousands and thousands of hours of content. You know, it's there's you know, where where the human effort needs assistance to sift through and understand audio and video at huge scale is really where we shine. And so I'd say it's mostly efficiency gain and cost savings, right, where we do for larger institutions, but also new revenue growth.
Ryan Steelberg:What if I take the Masters golf tournament, you know, they've been a customer. We've been helping ingest and index and understand their programming for years, but we're also because we've indexed it so well, we now help them license their footage faster. So when you see commercials that pop up, like you're watching the masters and then very next commercial break, you're seeing that same player show up in a Rolex ad. Right? That's being facilitated because of Veritone.
Ryan Steelberg:So so I'd say for media and entertainment specifically, we are not just a cost center that brings more efficiency to them, but I would say it's a net positive. Not only we're bringing more efficiency, but we're helping them generate more revenue and deliver faster ROI to their advertisers, which in effect, it's almost paying for itself very, very quickly. So again, to answer that question is I think we're bringing more utility and more revenue opportunity value to larger enterprises first.
Matt Allen:Yeah. That makes some that that example with a master's golfer that you just used makes a lot of sense. And do you do you guys do things with in terms of, like, let's just keep on sports is that if if player a is having a great game and then when he goes to that highlight at the end of the commercial break, is that Veritone?
Ryan Steelberg:A 100%. If actually, if you guys watch this for credit, if you ever watch SportsCenter, you're you are usually in the credits every single night at the end of the SportsCenter show. Right? So again, SportsCenter's probably the most interesting is because, you know, the master's content that you see is mostly their own content. What you know?
Ryan Steelberg:And to be clear, it's not just the content that you and I watch on channel two, right, on c v CBS or on ESPN. But every night, as you know, when they're creating like sports center, they have to sift through thousands of hours of content from coming from all different sources. We help do that. We help ingest that and make that very, very efficiently and fast. Yeah.
Ryan Steelberg:So that was that's a good example.
Matt Allen:And then kinda going I know you guys are getting involved in the public space and you and you mentioned earlier the the part about, cops. So can you kinda talk about that a little bit? And then also other other areas? I know you've worked with the air force, other public companies that you're or public enterprises that you're working with.
Ryan Steelberg:So I'll make this transition. In the world of media and entertainment, their main product is their same data. Right? The programming that they produce, right, is their main product. When you look at a police agency, they now have to deal with tons of audio and video, but frankly, it's like a foreign asset to them.
Ryan Steelberg:It's not their main product. It's not like they've been around for years becoming experts of how to like search through audio and video. So meaning the value that we're bringing to the public sector and government, state and local law enforcement, in my opinion, is even higher. Right? There is not a case anymore, or very rarely is there a case where you're not having to sift through hundreds, if not thousands of hours of Ring camera footage, you know, citizen upload videos, security cameras across the city.
Ryan Steelberg:You know, again, we're trying to not hopefully get to the point where we're slowing down or preventing crime, but the speed of having to ingest evidence and sift through that fast, this is a game changer, right? How quickly can we find the bad guy, right? And that's something that we're taking a lot of pride in. And I like to say is, hey, we, you know, the same value that we're bringing, the same improvements to our technology that is making ESPN run more efficiently is now helping catch bad guys faster, right? And so again, we want our police officers to be not wasting or spending all their time dealing with digital evidence, right?
Ryan Steelberg:We wanna be able to make that fast and get them the information and the intelligence so they can go find the bad guy and get back on the streets and, you know, and do their job. And so, but again, the important thing for this conversation and investors is same technology stack. Right? Our underlying technology stack, which we call aiWARE, is the exact same technology stack that's running the air force. Right?
Ryan Steelberg:Empowering a lot of their investigative work and sucking in drone footage, etcetera. But it's the same technology that's powering ESPN, and it's the same technology stack running Beverly Hills Police Department. Right? So just think of that as in terms of how efficient is that from an operating model perspective that I can get that much leverage out of the same technology stack.
Matt Allen:Now, do you have to get independent contracts with each police department or how does that kind of work?
Ryan Steelberg:It varies. Usually it's direct. So you have to go, you know, and through RFP processes, depending on what the price of your solution is. It's a great question. You know, we've, you know, our solution, you don't have to buy all the different applications or all the different use cases.
Ryan Steelberg:So we, so we, so for example, we have one product called Redact, which does AI based programmatic redaction. Ironically, we're not trying to find necessarily the bad guy in video. We're trying to protect our faces. You know, when you always see those videos get released to the public and it blurs out people's faces or somebody's credit card or their computer screen, that takes a tremendous amount of time too. Our software, our product called Redact, automates that.
Ryan Steelberg:And so for police agencies, they can buy just that application to start. So meaning they can get in under a budget constraint. So it doesn't have to necessarily go to city council to start the relationship. And then eventually, when they have more budget, can buy all of our applications. They can empower homicide and car detection and police detection and all those things.
Ryan Steelberg:I'm sorry, a person of interest tracking. So again, we usually sell direct, and that's building relationships with a sales force or other channel partners with lieutenants and chiefs and sheriffs, and you kind of start that relationship. So it's a very relationship industry, I would say even more so than the commercial Commercial, yeah. Media and entertainment side.
Matt Allen:And now do you charge per is it just a set monthly fee? Is it per query? Or how does that kinda work as well?
Ryan Steelberg:It's mostly not the amount of data that they're gonna be using. So we gotta keep it simple, right? To your point, we call it shirt sizes, right? Are they gonna be doing four gigabytes of data processing a day? You have to translate that, right?
Ryan Steelberg:And so typically you can do an evaluation working with an agent and do an assessment. How much data have they been ingesting historically? Or frankly, they haven't been able to do a lot of ingestion because they don't have the technology to go do it. So I think first is at least the cost variables that we try to turn into what a what an what an easy to understand pricing model is is fundamentally how much data they're to be ingesting and what app ultimate end applications are they going to be using from Veritone, right, to turn into value, like redaction software? So, it's two main variables.
Ryan Steelberg:What application, and really how much data are they gonna be ingesting and processing?
Matt Allen:Now, you mentioned Ring earlier, just when you're talking about local law enforcement. Is Veritone partnered with Ring, or is it?
Ryan Steelberg:No, it's not a form of partnership, but they have access points where, in the context of an investigation, typically, you talk about citizen upload you know, footage Mhmm. They have a subpoena or some court order that you know? So that gives them the authority to go collect the evidence. Our software makes it easy for them to make those ingestions. So the so the vision here is think of us, Baritone, as an open platform.
Ryan Steelberg:We don't care what type of data source it is. We wanna make it easy to get it into the platform, right, ingest it. There's there's how many how many thousands of different camera providers are there out there? Right? Tons.
Ryan Steelberg:Yeah. Tons. So the key here is, no. I don't care what type of data it is. We wanna be able to ingest it and then use our AI to, I'll call normalize it so we can look at it like homogeneously, right?
Ryan Steelberg:I like to say as crime travels, right, they bounce between different cameras, right? God forbid, you know? Yes. So therefore you have to have a solution, right, that can look at and interpret any of those data feeds. And that's kind of
Matt Allen:what we built. Yep. That's awesome. Do you see yourself just keep on doubling down in the public sector space? Is that the priority or is it just kinda fiftyfifty?
Ryan Steelberg:I think it's kinda fiftyfifty. You know, we have such a, I mean, I think we're one of the rare companies out there that has a pretty healthy diversified portfolio between, again, commercial and public sector. Also, the contracts are bigger on a per unit basis in the Fed at times, but harder to predict when those close. So it gives us the ability to hedge a little bit as a business where I have my more repeatable business in kind of state and local and my more repeatable business in commercial, and then it affords us the ability to go after those bigger deals in Fed, like the Air Force, right? That can take years to go land.
Ryan Steelberg:I think that's a huge opportunity and it's something that makes Veritone very, very unique. One new application though came kind of organically emerged, which frankly, just a couple years ago, we didn't really know this was a great opportunity for Veritone is we're in the AI model training business now. You know, because we've been ingesting for ten years now, like millions and millions of hours of audio and video and preparing it, we are now working with our clients, and we're turning now that audio and video into AI model training data and and actually licensing and selling that to the biggest hyperscaler names you know. So when you think about the next generation ChatGBT or Sora three, etcetera, Those take lots and lots of training data to create. Veritone is now a key vendor.
Ryan Steelberg:We're a key player in that space focusing on mobilizing those audio and video data sets.
Matt Allen:Yeah. That's what I was gonna get to next is your new venture into hyperscalers. How does that work in the sense of, let's just use Disney and ESPN as an example, are you selling that on their behalf? Yes. Or you're just gathering them?
Matt Allen:Okay, so that's kind of how it works?
Ryan Steelberg:Exactly. So in effect, think of it as they have a data, we're not limited to their data sets, but obviously, you know, when we are now helping one of the big AI, you know, hyperscalers, let's just say hypothetically a company like Meta, okay, who is looking to train a new model, if we have the relationship, if we already have the data from our customer, great, we can do it pretty quickly, right? And in effect, helping, we have to obviously tokenize the data into the format that they need, and that's additional value add we bring to the table. We actually are creating the labeling or the tokenization of the data
Matt Allen:Yeah, through do you mean by tokenization of it?
Ryan Steelberg:Tokenization is a term you're gonna hear a lot more about, try to keep it simple, you know, when when you convert audio and video into bite sized data chunks or or or digital representations of what's inside the content, you can call it metadata. That's more of a classic older term. But when you the definitions or the representations of that video, when you have to load it into AI model to train, you call those data elements tokens. So very similar to when you read all the stuff about these large language models of how many tokens it took to train the model, that's what we do. We help convert unstructured audio and video into discrete, I'll call packetized tokens that then can be interpreted or used by the data science team to train their models.
Ryan Steelberg:You can't just give raw audio and video, right, to a lot of these machines or these training platforms. You need to give them the the actual audio and video file tokenized so they can understand what they're training on, if that makes sense. So think of it as just a way of digitizing or it's a way of expressing the audio and video that can be read and ingested and understood into these AI models that are being trained.
Matt Allen:That's incredibly interesting. I know the market there is absolutely huge, especially right now. Kind of from more of an investment standpoint, I'm personally wondering, that how does Veritone get paid for this? Is it as a referral fee? I know you're work I know I said you're selling the video, but does does say company A who's partnered with you, say a Disney, pay you or does the Meta pay you or No.
Ryan Steelberg:We collect. We're we're the prime. No. So this is not a referral. We do a revenue.
Ryan Steelberg:So we do pretty much all the work. Right? So we like we're already doing. We're ingesting all the data. We're we're tokenizing it.
Ryan Steelberg:We're the ones dealing directly with the hyperscalers. We're we're the ones licensing it. We're actually working with their data science teams directly, and we set price to be very clear. So we kind of do everything, and then we do a rev share back to, and that's the most dominant or custom thing. So, we account for gross, we're in control, and then we do a revenue share back to the suppliers.
Ryan Steelberg:Yeah, I
Matt Allen:mean, that's huge, in my opinion, huge on multiple angles for Veritone and even the client because, you know, they're making money doing nothing for a service. So, I think that's awesome.
Ryan Steelberg:I just, I mean, you just touched on it. I mean, we're excited about it because it's just another form of them being able to monetize their content. Right? You know? And and that's our job.
Ryan Steelberg:Our job is, you know, I don't want the masters to ever leave. Why, like, why would they? If, you know, like, we work with them. We get they get utility value. Right?
Ryan Steelberg:Because we're indexing their content. They're using it for their own internal purposes. We're helping them license their footage, now we're helping yet create a brand new revenue stream, right, by helping them package and license in a very secure, right, legitimate, you know, methodology to these hyperscalers. So it's I will say a great way to think of it is another moat, right? It's another reason why we can, you know, we have confidence that we're gonna re sign CBS, we're gonna re sign ESPN over a longer term period of time.
Matt Allen:Yeah, that's awesome. Kind of, I have a question for you as someone who's always been on the cutting edge of technology. Where do you see AI moving in the next five to ten years? And not in a general sense, but let's just say with content and media. How do you see it really taking over?
Matt Allen:For example, I know YouTube wants to enable creators to allow them to create documentaries using I mean,
Ryan Steelberg:Yeah, do mean, it's exciting. You know, living here in Orange County, California, you know, I have a lot of family and personal friends who are in the biz. My namesake cousin, Eric Stielberg, you know, is Jason Reitman's best friend. He shoots the Ghostbuster movie. So it's part of the family.
Ryan Steelberg:We have lots of writers. I mean, this is impacting every aspect of content production, period. You know, it's like CGI back in the, you know, the mid early eighties, what it did for filmmaking. I do think it's gonna be You're gonna see some macro trends is, Matt, if you have a great idea, you're gonna be able to produce a quality, full length feature movie, Right? Just out of your head.
Ryan Steelberg:And some people were like, that's ridiculous. That's a travesty. I'm like, but is it? You know, if you have the best if you're a great storyteller and you have a great idea, should you not be able to communicate that in that format because you can't write effectively per se, or you can't afford to book out a studio and produce it? So I think it's gonna be messy.
Ryan Steelberg:The rest of the ecosystem's gonna have to figure out how the jobs work and how the ecosystems work. But I mean, I know for a fact that I'm gonna now, I'm gonna see or listen to something over the next few years that I never dreamed about. Right? Like the new avatar is gonna come out, it's not necessarily gonna come from a major studio. It's gonna come from Matt Allen producing it.
Ryan Steelberg:I think the thing I would say is hyper personalization. Mean, right now we're still building one for many, and I think you're gonna get to a point where it's gonna get really cost effective, where there's gonna be hyper personalization of not short form, but longer form and long form content for everybody. And I think the lines are gonna blur between AR, it's already happened, but I think you're gonna see a rebirth. There's been a few starts and stops of augmented reality and virtual reality. I think that that's gonna come back into the fray as well.
Ryan Steelberg:So, know, it's gonna be very disruptive, and I think it's gonna force a lot of traditional production based and Hollywood like models to change, but I do think that there's no putting the genie back in the bottle, and I do think it's gonna be greatly empowering for the creator, the influencer, and you're gonna see a lot more personalized content. I think monetization is gonna have to catch up as well. It's Right? Gonna be harder. I think you mean you're gonna have programming in theory that's gonna be one to one.
Ryan Steelberg:Think of it a movie that's created just for me, right? And so what does that mean from a business model perspective?
Matt Allen:Well, that's what I gonna ask, what do you mean hyper personalization? Do you mean that it can mean that when you're scrolling through your Instagram feed or YouTube feed that something's created instantly for you?
Ryan Steelberg:In narrow time, yes. I think it's, you know, yeah. It's gonna take a little bit more time where stuff is being generated on the fly, but you know we're gonna get there, right? Just ad creative is being assembled in near real time on the fly today, you know, content, it's gonna start short form, then long form, I mean, then medium form, etcetera. But yes, 100% that is gonna happen on the fly.
Matt Allen:And do you believe that's gonna be through AI agents? And do you think that comes through companies themselves or through, do you think there's gonna be a push and pull where companies say, hey, we can do this better than in creators, or do you think that will allow the creator to-
Ryan Steelberg:Yeah, it's gonna be a battle between push and pull, right? You know, again, you know, we've gone through several times where I'll call, you know, the vertically integrated platforms. They are important sometimes to bring some continuity to advance something, right? And then they kind of get oversaturated or they become monolithic, and then you see competitors kind of break in, right? AOL was a great example, right?
Ryan Steelberg:I mean, again, whatever we think about it, know, AOL wasn't the open web back in the day, as we all know, but because of the simplicity that they brought to it and they owned it every end, right? They owned the data and a lot of the application, a lot of the content in there, it grew. I mean, the people bought modems and stuff like that to get into the game because of them. But over time, as the technology, people said, no, I have a better program. I have a better content.
Ryan Steelberg:My website is better. You don't need to go through the walled garden world of AOL. Come over to the open web. So I do think this is gonna ebb and flow. YouTube is such a major player.
Ryan Steelberg:Right? They've, you know, they've been they've they've created an and and and again, just like everybody in Microsoft in the past, if they see a tool or a startup that's created something incredibly new or novel, they're they're either probably gonna buy it or build an extent. It's just their MO. So, again, I think the balance is here right now. I feel we're weighted in a little period of time where it seems like our entire ecosystem is dominated by eight companies.
Ryan Steelberg:Right? And whether it's Nvidia with certain content players and OpenAI and you're seeing that, but like every time I do think it's gonna, equilibrium will come into the fray, horizontal complementors will break in because it's happened over and over and over again. And I think the same thing is, frankly, specifically for the creator and the influencer economy. The same thing's gonna happen, Matt. There's gonna be, you know you know, frankly, lot of the tools that YouTube is incorporating now into their direct tool set used to be third party companies.
Ryan Steelberg:Right? And now they brought a lot of those technologies natively into YouTube. So it's gonna ebb and flow.
Matt Allen:Yep. I completely agree with you on most everything you said. One thing that I even probably agree with you, even double down on is that I think you'll have micro entrepreneurs as well in the sense of, you see a lot with, I'm 31, I'm kind of on the millennial gen Z line, as I call it. Yep. Yep.
Matt Allen:But a lot of these gen Zers are creating SaaS companies using vibe coding and things like that, where, hey, maybe you don't have a job that they could have got $75,000 but I see them moving where we see it have an economy where solo entrepreneur gets four or five local businesses for their certain SaaS, makes $75,000 a year, and that's kind of like their new job. And so I see a lot of that kind of happening.
Ryan Steelberg:You know, it's happening right now. I mean, I don't think there's, you know, 80% of every college junior and senior, I've heard some crazy study that is doing some form of prompt based vibe coding to come, you know, to explore. And they're they're they're probably POC level applications and workflows, but that's just, you know, the game has changed. You know, I do I do think so. I think for for for corporations like us, you know, we gotta figure out, like, who's who's paying our employees bills.
Ryan Steelberg:Right? Am I gonna be paying healthcare if they're gonna be working on four other projects? I do think because of the tooling has made it easier for non engineers, for the most part, start getting into and building workflows and applications very, very quickly, I think there's gonna be this trade off where, hey, people wanna make a good quality living. They may need to be running three jobs at the same time, but what does that mean for Veritone? Do I have to pay for all their healthcare if they're gonna be working on three other things?
Ryan Steelberg:Yeah. So again, I think you're touching on something that a lot of people I don't hear enough of is, you know, this AI phenomena is, you know, people are mostly thinking that I'm a single job person, you know, that dominates our historical, you know, society, I think it's gonna change, right, over time. You're gonna see entrepreneurs that are gonna be running multiple projects and multiple companies at the same time, but they can still do it cost effectively and efficiently.
Matt Allen:Sure, I completely agree. And that kind of leads me to kind of just had a thought about you being an ad tech expert historically. Yep. Do you kinda think of the highly personalized angle of it all? Do you think this is gonna be the peak of ad tech, is that you're gonna have the most highly personalized ads that we've ever had?
Matt Allen:And do you see that something that you might get into either with Veritone or something else down the line?
Ryan Steelberg:You know, ironically, we got out of the ad tech kind of space. Like we used to, you know, I, you know, we actually owned an ad agency, right, and we powered a lot of the differentiation through it. And just like everything, that was more of a service based, consumption based model versus SaaS. So I think we'll probably stay in the game where we wanna help people who are, like ESPN, we're trying to help them make their sponsorship and advertise programs better. Right?
Ryan Steelberg:Yeah. I don't think we're gonna be back in that field on a first person basis, but we are there already. Excuse me. Excuse me. On a secondary basis, we're there.
Ryan Steelberg:The sweet spot of, know, yes, we're very quick to respond. When stuff gets a little too personal, we, you know, kind of self regulate. You know, the kind of joke is like, you're having a conversation about, you know, a new type of wine that you're enjoying with your wife, and in three minutes you're seeing an ad for the same wine. Whether it's targeting, now you take that same level of real time targeting, and you do that with dynamic content creation with AI, So I think there's always a balance, Matt, that, you know, we will accept and then push back on. You know, I'll give you a stat.
Ryan Steelberg:I don't even know what it is today, but in the, like, around 2009, it was estimated that the average American was seeing, like, over 4,000 unique ads a day, and just a decade before that, it was like 152. So that's the other thing is is much time, how much, I mean, just sheerly, how much can we process and retain? Right? There's a diminishing return, right, always. So the answer is the technology, I think, Matt, going to outpace it, but I think there's, like for everybody, what your personal load balance is, is gonna be radically different than mine.
Ryan Steelberg:Right? And if you're trying to get me engaged, you may have to serve Ryan one fifty ads, even though the engagement would maybe higher, and Matt, you know, a lot of people are like, no, Matt, I have to hit you over the head 10 times, right, with the same ad before you engage. So I I again, if they if they if they do it right, I think that to your point, the tools just keep getting better and better, so it could be in theory, the sweet spot for ad tech, but at the same time, you know, very quickly is, you know, let's just make sure that it doesn't cross over and there's consumer burnout very, very quickly because of it.
Matt Allen:Yeah, for sure. And just for the people who are listening and kinda what I meant by that question is is that when Ryan was saying, hey, we're gonna have highly personalized content, some point down the road, which means, hey, if you're scrolling on YouTube or Instagram, that a video for yourself that you're gonna be find that you're gonna like a lot, it's almost instantly created. And my thought process there is that, you know, you have an ad that where, hey, we know you really like these type of ads, you engage these types of ads. Can you do it instantaneously? And that's down the road, but it's, I could see that sort of landscape moving forward.
Matt Allen:It's awesome that Veritone's part of that as a secondary.
Ryan Steelberg:Yep.
Matt Allen:What do you think is gonna be the biggest industry that gets disrupted by AI as Ryan Stilberg, the person, not the CEO of Veritone? Just
Ryan Steelberg:Oh, as a yeah. As an individual?
Matt Allen:Like, the hottest take that someone's not talking about right Yeah.
Ryan Steelberg:Mean, you know, there's so many different flavors of AI. I do think robotics. Right? I think it's gonna come much faster than people think. You know, I I and how we experience robotics, you know, are you gonna have, like, a humanoid walking around your house, or is it gonna be, you know, somehow, you know, an army of micro drones that do little different things?
Ryan Steelberg:I just think that there's gonna be an interesting crossover more into the physical world. I think, obviously, we've talked a lot about it. Media, entertainment, and content, production and consumption is gonna be radically disrupted, and coding itself, I mean, again, you're pretty attuned to what's going on, but if if you're if you're touching a keyboard historically to do something, probably gonna get disrupted. Right? And if you take and if you take, you know, tactile using a keyboard to do something, a, and what you are trying to produce is very structured like code, It shouldn't be shocking to anybody that some of these larger Fortune 500 companies are estimated that 30 to 40% of their actual code writing is gonna be done by AI.
Ryan Steelberg:Right? So again, I think my prediction is some people think that true robotic as part of day to day in our life is gonna be like fourteen and fifteen years away. I think it's gonna be seven. So I think it's gonna come a lot faster. And again, I think we're seeing it right now, we're talking about it now, is the speed and advancement of content creation and distribution is happening so fast that, again, we're seeing it in real time.
Ryan Steelberg:I just think it, may jeez. Every six months, the quality, right, of these models and engines for content creation keeps getting better and better and better and better. So I would say, you know, those two are ones that I just I I feel I have a lot more understanding and intimacy with the content side, but I just just knowing knowing how they're knowing what knowing the type of training data they're trying to get their hands on, right, of, digitizing the real world in terms of three d emotion, I just feel it's gonna happen. It's gonna empower robotics a lot faster than we think.
Matt Allen:Yeah, and if you're listening at home is that one thing with the commercial robot, humanoid robotics, is that we have around a million person shortage in labor, meaning that there's just people who are in the age of 21 to 40, but just don't want to do physical labor jobs anymore. And, know, that's kind of where society is going. I hope society keeps on developing it towards that, which means that humanoid robots are coming on very quickly for a commercial. And I think, as you said, once that hits, it's gonna be night and day. And that kind of leads me to the guest question of the day from actually one of my best friends who's a big fan of you.
Matt Allen:His name's Tim Anders. He lives in Tallahassee with me, and he wanted me to ask you this. How can you ensure inference or any type of data that you guys are giving to large LEGOs models to train on remain free of bias without using regulation? That's kind of his, that's kind of his, he said, you can't say regulation. Yeah.
Ryan Steelberg:Okay. So the type okay. So first is the the content that I am packaging and selling. Let's take let's say it's sports content. First is, are are is there an assumption that the content before I turn into tokens is biased already?
Ryan Steelberg:K. So if I'm taking footage from a basketball game. Right? And I'm not trying to skirt the conversation or the question, but my job, what we get paid for is I'm taking that content, tokenizing it, and they're and they're using that to train. So by default, if the content, original content itself, in theory, is biased, turn it into training data, in theory, would carry that bias.
Ryan Steelberg:That make So I don't feel what we're doing from a lot of these things is is introducing or creating a bias. Now I'll give you I'll go I'll go I'll I'll bite on that. I'll go a little deeper. Part so some of some of the the training data, let's say, is used at to help recreate the ideal swing in golf. What does that even mean?
Ryan Steelberg:Right? The ideal swing in golf. Like, for example, they necessarily need to use professional footage of Tiger Woods to train a model. Right? They can video me swinging a club, right, if I was right handed.
Ryan Steelberg:But the but the question is, do they if they're trying to create a simulation for generative AI because you wanna, right, create content that has somebody swinging a golf club, do you want that swing to look like Jack Nicklaus or Tiger Woods, or do you want it to look like Ryan Stielberg, the 16 handicapper? What I just described there, some people have strong opinions on. Right? That in theory could be a bias. I think more acutely is, in some areas that we do do business with, it's not really for model training, but it's derivative is, you know, we are involved in hiring.
Ryan Steelberg:Right? And this is where a lot of conversation about bias is coming into is if somebody are applying for jobs and and and and the company is using an AI based bot system, natural language, you know, platform chatbot to communicate with applicants. Right? You know, are they trained on a bias? Do they have, you know, some racial profile or skew in there that may prevent it?
Ryan Steelberg:Those are areas that 100%, I think we proactively, there's not a ton of laws. Actually, in New York, is. We actually went through a pretty laborious bias analysis on what we do in applying our AI to programmatic hiring, is to go through and make sure there is no bias, right? That I can't really discern, or even if I could discern or interpret or make an assumption based on how you're speaking, right? Am I saying you're from the Deep South, for example?
Ryan Steelberg:So again, I think there's areas where, know, do I feel that I'm introducing or need to evaluate the potential bias by doing, helping models train for sports movement? No. I think like anything is, there be an awareness or evaluation of bias depending on different applications of AI when I'm training? 100%. Okay.
Ryan Steelberg:So anyway, think, but right now, I think it has to start first. The last thing we want is every state and every county and every city to have a different policy, right? You know, this is where I do think that there needs to be more standards created, and God, we're gonna adjudicate and fight that out forever, but in the meantime, companies like Veritone and others, it's common sense in my mind, right? We gotta self regulate first until the legal system and legislation can catch up.
Matt Allen:For sure, yeah. That's awesome, and that's an issue, I think, we're dealing with it now slowly, but it's just gonna keep on, like you said, it's gonna keep on coming more and more and more in his life and in our life. And so we, me and him were having lunch on Friday and he was like, Hey, you gotta ask him this. But no, first of all, thank you so much for your time, I know you guys just had earnings, and where's the best place where listeners and viewers can go and learn more about you and follow you and see some of your content and what's going on with Veritone?
Ryan Steelberg:I mean, veritone.com, just go to our website. It's great, I mean, very quick, you could say, hey, what are they doing? We have industries at the top. You can see it very easily and learn and watch videos of what we're doing in public sector versus commercial. You can get to our investor site from there.
Ryan Steelberg:We did have a great quarter. We had to take a non cash charge from an asset that we sold over a year ago that kind of the bots picked it up and somehow it was misconstrued that we lost more this year than we did last year, but let me tell your investors, no, we crushed this quarter. Okay. Our business is running much more efficiently, And again, you can learn more about and everything else from our website, but that's a great way to start.
Matt Allen:Okay. Well, perfect. Like I said, thank you again for taking the time, I look forward to following you and follow Veritone's journey for sure.
Ryan Steelberg:Thanks, Matt. Thank you, Tyson. Absolutely. Appreciate it.