Introduction
Ryan: Welcome everyone to the AiFounders Podcast Show. Our podcast is dedicated to celebrating the remarkable accomplishments of AI innovators, entrepreneurs, and visionary founders and the captivating stories behind the movements they’ve built. I’m your host, Ryan Davies, and I have the honor of hosting today’s episode, Transforming Media and Entertainment Through AI Video Innovation, with our special guest, Sam Bogoch. Sam, thank you so much for joining us here today.
Sam: Great to be here.
Ryan: This is going to be a really fun topic. We’re going to talk a lot about, again, just like what AI is doing, right? How are we revolutionizing everything? How can businesses really take advantage of this? How can they accelerate their media and entertainment, as well as their impact on video production, personalization, and targeting? So this is going to be a great episode for our audience to listen to and be able to take a lot of this and apply it directly. For our audience, I want to talk a little bit about Sam before we get started here. He is the co-founder and CEO of Axle AI. Give it a check out after this episode and learn more. Really, it’s where software combines the power of AI and machine learning, either on-premise or on the cloud, and brings this cross-platform video workflow automator and collaborative editing platform, making it easy for everybody to be able to take advantage of their media and entertainment, videos and clips and everything, bring it together. We’ll talk more about that. He’s advised a number of tech startups and is the co-chair of the Boston Chapter of the Harvard Alumni Entrepreneurs Association. Really is a specialist in media asset management, server architectures, customer interactions, everything that we love hearing a lot about here on this show. Before this, he spent a number of years in director-level product design, product management, and business development roles at Avid, where he helped grow from a very early 1.1 version of it into an industry leader and household name with Axle. Sam is bringing years of media management experience to an emerging market of teams around the world in need of simple tools to manage their media. Sam, no one wants to hear me talk anymore. They’re all here to listen to you. So, I’m going to turn it over to you if you want to give us a brief overview of your background and how you became the CEO of Axel.ai.
Sam: Thanks so much for the buildup. I appreciate it. Sounds good. I’m up for that. So yeah, essentially, Axle was born of the realization that, first of all, there were a lot of unmet needs among video creation and content creation teams in terms of being able to search and repurpose their video. Anybody who shoots video knows that it just piles up very quickly often terabytes, tens of terabytes, turn into hundreds of terabytes, and in some cases, petabytes. We have a growing number of our customers that have petabytes, which are thousands of terabytes of content, and it extends. It used to be that only big broadcasters and movie studios had that much stuff, but increasingly, people like universities, churches, sports teams, and governments are all shooting a ton of video, and it ranges everything from training videos to reality shows to movies, and it’s not all where you expect to find it these days. That was the big unmet need, and then on the other hand, you saw AI starting in about 2018 when we were founded, starting to become relevant, starting to be able to actually analyze the contents of the video. So that the problem was always you could search your stuff, but you would need an army of interns, and it just wasn’t loggers, and nobody could afford to hire them, and it just didn’t make any sense. There were a few cases where it did. I’ll never forget. I visited Entertainment Tonight back in the day when they would do kind of saturation coverage of the Oscars and the Emmys and so forth. On red carpet night, they would literally have a squad of people all typing furiously from different camera angles. George Clooney looks left, George Clooney looks right, George Clooney is wearing Dolce and Gabbana shoes, and they were just like a squad of these people, but only Entertainment Tonight, or someone like that could afford to apply that many people to a fairly basic problem like that but with AI, it’s kind of widely available now. Anybody can get their stuff analyzed in theory. The hard part is doing it in practice. The technology exists today, but bringing it to where the media is and making it cost-effective and user-friendly. So you could just say, hey, I need that footage of George Clooney, and the shoes, the brown shoes. So you want to be able to just type in George Clooney and brown shoes and find it, and that’s what our software lets you do.
The Dominance of Video in Today’s Landscape
Ryan: That’s incredible. One of my very closest friends is the video editor. He’s the multimedia manager for the NHL franchise right where I live. So, man, he does so much work in this space, and I’ve been to his office before, and I’ve seen some of the videos. I mean, he does incredible work. He works with the NHL and things like that and the amount, like you said, the number of clips, the mountains of, like, he’s always talking about, yeah, well, I’ve got this hard drive, that hard drive, this, to the cloud. It’s unbelievable the amount of stuff that he has, right? So, like, I could totally understand it and grasp that concept of all of a sudden, I want this skater firing a slap shot against the Bruins or whatever, right? Like against that team sort of a thing. So being able to pull these clips together and do it, is such an amazing ability to do that. What a great mission being solved there. Let’s kind of break down a little bit more and go a little bit back here, talking about video media and entertainment from your perspective. What kind of role does it play in the current landscape? I think video is still the dominant way that we love to consume content. I mean, we saw it this week with the announcement, Twitter, I guess, X announcing that they’re a video-first platform going forward. So, I’d love to get your take on that.
Sam: Yeah, well, it has been building and building over the last couple of decades, and I’m old enough to remember when the video was still pretty exotic, and you could get it over the television. You could get crummy versions of it streaming. But you know, even something as basic as an HD camera was like a big piece of equipment that only the biggest networks could afford to buy. Never mind, 4K, it didn’t exist. So what’s happened, of course, is it’s all been democratized, and everybody has a smartphone. Everybody has a 4K camera in their pocket. So, the potential to shoot a huge amount of material is there, and increasingly, it’s happening. Then it gets uploaded to YouTube or social media, but because of the increasing familiarity with shooting this stuff and the bandwidth available for uploading, it’s the same starting to take over from still images for sure, and definitely from text. So we can definitely, I think that’s an accurate statement that the world is kind of video first today, but it probably was not the case five years ago, and definitely was not the case 10 years ago because of all these limitations. No one knew how to shoot good videos. No one knew where to store it. You would instantly fill up your phone or your laptop computer. I don’t want to say those are totally solved problems today, and of course, the transition to 4K and now 8K is just making it worse, right? So the bigger the footage, you’re sort of back to the future, and you have the storage problems and the transmission problems, but the video is there, everybody has a ton of it, and everybody knows how to play and communicate with it, let’s say. So, yeah, I think that’s kind of the way things are headed.
Challenges in Managing and Utilizing Video Content
Ryan: Absolutely. I remember even not too many generations ago on my cell phone, right? It was that, well, you could take 850 more pictures, or you can take six minutes of video, and then you’re, that’s all we’ve got for you. You’re like, okay, well, I guess I’m not taking as much video as I want to. Right. Everything was, like you said, still photos, and that’s how we kind of logged everything, and we’d love to take video, but it was, it was cumbersome. As you said, storage was an issue, arranging it was an issue, and going back to find it later. So this is all fantastic from that perspective. We talked about it a little bit there. Some of the common challenges that companies face when it comes to managing and utilizing video content effectively, there’s a little bit of an archaic nature behind it previously, but talk about the ways that whether either Axle.ai or just other AI-driven solutions that you’ve seen really helped to Overcome the challenges and maybe if there were more challenges to shine a little bit of light on those for our audience as well.
Sam: Absolutely. Well, the second half of our story is the AI, right? I mean, of course, we’re about video, but we’re video using AI, and that’s the other thing that’s completely blown up in the last year. I don’t have to tell you, which is that AI is the number one story in the world. They had Davos last week, as well as all the CEOs and all the heads of state. I mean, yes, they were talking about the Israelis and Hamas and Ukraine and the Russians. Sure, but AI was like the big story to us, and everyone would get into them, come up to the interviews, and everyone would be like, I just want to talk about what’s going to happen in the next 12 months with AI. Even companies and countries where you didn’t think it was going to be the number one topic. So, it is unbelievably powerful and ubiquitous right now. First of all, Generative AI, which was the big buzzword of the last year, is now coming into its own as regards video. So, most of the first wave of Gen AI applications were text-based. If you look at the strengths of ChatGPT or Llama or these other first-generation models, they were really good at basically bullshitting, right? Pardon my French, but generating huge amounts of text that sounded good and might or might not have real facts in it, but it was like, wow, that’s great. Now they’re applying enough computing power, and they have big enough data sets that the focus is going to be increasingly on mixed media and essentially the ability to converge, models, visuals, auditory, and text worlds into a single unified model That can be used for creative purposes. So you can increasingly you can tell these LLMs. I want a 30-second clip of this person doing this, and there it goes. You can also do completely synthetic things that didn’t really exist before and couldn’t have existed before, but you could also use it to refine and identify the important parts of the video that already exist, and obviously, that’s a little bit more. What’s the right word? It’s a little bit safer. It’s a little bit better understood. It’s kind of the Wild West with these LLM models, and a lot of the sample videos that you see coming out there are wild-looking, and they look like VFX reels, but they may not actually connect with people or have content that is of value. Whereas if you have a whole lot of legacy footage, let’s say you’re in the sports business and you mentioned the NHL example, you may have a ton of footage. In fact, the NHL did a lot of the pioneering work in this area where they were analyzing all of their historical media for certain things happening. They wanted to get statistics, better statistics. I remember, gosh, I was talking to one of their top scientists six or seven years ago. They were analyzing all of the footage they had for the particular hand signal that a referee makes when a goal is disallowed. They just wanted to figure out how many times over the history of the NHL, goals had been disallowed. I don’t know, something with the elbow. So they were able to analyze that going way, way back. Now, again, that was a very specialized use case, but it’s only one of many examples of how good information is buried in years and years of footage. More often, it’s things like political campaigns or corporate marketers. We need a clip of our CEO talking about sustainability because he’s going to be on Bloomberg in two hours and we need that like eight-second highlight clip. So they’re rushing around. In the old days, there was no chance of them finding it, and now, with our software, you can literally say you’re looking for the CEO saying sustainability or some other word, and because it combines, it’s kind of a synthesis of speech transcription, face recognition, objects, logos, and scene understanding. These are all in our model if you will. You’re able to search across all of those as if you were searching Google, but you have an absolute ton of material, if you’re one of these customers, to mine with it. So you’re more like mining the material that you have rather than with Gen AI generating brand new stuff from scratch.
Future Trends and Technologies
Ryan: I love that, and I know, like, I’d like to drill down on that a little bit more ’cause we’re, we’re like, really in the era of personalized content, right? The more personalized, the more targeted you can make it, the more effective it is, and the more impactful it is around that side of things. So, I mean, maybe just speak a little bit more about that and how AI is really enabling content creators to tailor their videos to these specific audiences and still not have that massive cause lot. There’d be a lot of overhead before, whether it’s actually capital or time or whatever it is to be able to do something like that and maybe not get the return on it, but now, this is really that next, that next evolution, right?
Sam: Absolutely, yeah, and increasingly, communication is narrowcast with social media in particular. It’s very hard to craft a single message that you can just blast out there, and everyone just has an aha moment, and you’re done. It’s never like that. You have to create highly targeted little snippets for each subgroup, and that’s equally true in the marketing sphere in politics. We’ve had candidates that have used our software for their town hall meetings, where they go out on the road and they’ll do these town hall meetings where essentially it’s like an AMA session. People can ask them anything for hours at a time. That generates hundreds of hours of footage, and potentially, you’ve got stuff in there to answer almost any question that comes in. So unlike a traditional campaign platform, let’s say, which would be like, I don’t know, a couple of pages of single-spaced text. This is more like a living, breathing thing, and someone comes on Facebook and says, I bet your candidate doesn’t even have an opinion about Somalia. Maybe they’re from Somalia, or they care a lot about Somalia. So that’s their hot topic, right? If you can quickly find the time in the town hall meetings where this candidate was asked about Somalia and mark in, mark out, push it out, that is a much more effective message than some kind of generic blah that’s designed for the widest possible dimension.
Ryan: I love that. I mean, when we’re in an election year, so I think we’re going to see it more and more taking place, right? With that happening or sorry, I guess you’re in an election year in the United States where you are, but the rest of the world’s watching very closely, obviously. So I mean, that’s just a wonderful analogy, and I think this is going to, again, just accelerate the use and the acceptance and how it’s going to go with that. How do you measure the impact of it, then? Like, you know? How can AI be leveraged to measure the impact and effectiveness of what people are trying to do with this content? Are there any specific metrics or analytic tools that really come into this space or come into play?
Sam: There are, and one of the things actually for driving that engagement is metadata, by the way. So the other big problem with video is that even if you can find the right video, put it in the right format, and post it on the right medium, no one may find it because of the associated metadata or lack of it, right? So, the other cool thing that AI does for you is that it bakes in all the metadata you’re going to need. You have the transcript. You have the name of the people, you have the brands, if any, that are displayed or mentioned, and so when you’re uploading to YouTube, you can cut and paste all that tagging information right in there, which means that a lot more people will find the content and it’s going to be stickier and more viral than if you didn’t have that metadata or had to kind of wing it. So that’s our portion of it. There are entire companies whose job it is to track the click-throughs, to find ways of measuring audience response, and to do emotive analysis. You can even, in some cases, get the feedback of the viewers from the cameras, see whether they’re smiling, frowning, or walking away altogether. That whole area is not what we do, but it’s kind of kitty-corner to what we do because we’re on the content creation side. Of course, the content consumption side is, if anything, even bigger.
Ryan: There’s definitely the intersection there, right, of being able to bring those together and seeing how they’re leveraged in the same medium in the same way. When we’re talking about kind of the future of this and bringing it together are there new emerging technologies and trends that you believe are going to continue to rapidly accelerate this? Have we hit a plateau, or is there a level of are we just at that kind of level of acceptance of this where people are learning about what’s even possible?
Sam: No, there are massive trends. This is all changing. I would even say month to month. It’s really very dynamic. Last year, the focus in AI was on these huge centralized LLMs. That was the only way to build them in 2022 and 2023. But this year, the focus is on edge computing and bringing the power of these massive LLMs down into small enough, manageable computing units that they can be deployed on a corporate network or even on a phone in some cases. So you’re starting to see some of this on the phones from Apple and Google. We’re kind of in the middle of that where, okay, we don’t need to shrink-wrap it down to your phone, but you need to have like one or two servers in your network that can do a lot of this AI processing for you, and the reason that’s important, interestingly, is partly driven by Generative AI. There was so much aggressive cloud scraping of other people’s data over the last few years, some of it without their knowledge or consent, that everyone is rightly paranoid now about putting their most valuable material in the cloud, and they also overshot, like during COVID, everybody had to go to the cloud because there was no going to the office. But now they have some choice in the matter. And they’re saying, wait a minute, why to park my stuff in the cloud where it could get scraped versus having it on my corporate network where it’s a little more secure? So that trend of kind of bringing things down on-premise, being able to run them at a reasonable cost, because the costs in the cloud are unbelievable, right? They’re sort of exponentially expensive the more resources you bring in, and what we’re finding is that it is possible. So our focus as a company is bringing all these cool cloud techniques down to where they’re affordable, easy to use, and kind of friendly for the individual video team but that’s not just us that’s doing that. I’d say it’s across the whole AI industry. It’s kind of a democratization of AI, the same way you had the democratization of video.
Ryan: So with that in mind, are you seeing that there are some adoption challenges? I mean, that was one from prior, like you said, that is the industry’s trying to overcome and really helping for the broader adoption of AI technologies in the field of content creation, video production, and video management. How can businesses overcome some of these barriers? But before we get to that overcome part, maybe I’ll let you shine some light on what you are seeing from any adoption challenges that are in play.
Sam: I’d say the biggest ones are that it’s almost happening too fast for people to get their heads around. The minute they get their head around something, 30, 60 days later, there’s some new thing that’s even cooler. And it’s like, where do I even start? We’ve had some very big broadcasters and production companies who have become our customers because they trust us to kind of shrink-wrap a lot of the cool new things that are happening for them and bring it into their world, but it is hard to keep up at this point. The approach that we’re increasingly taking is being a platform. Like we can’t possibly, you know, we’re a 23-person company. We can’t possibly bring together all the best developments of any conceivable AI-driven search for video. There’s just no way. We can do our piece of it, and then by staying open and by using open standards, we can bring in third parties and build kind of an ecosystem, and that’s what the customers, in the end, are going to want because they know that no one company is going to be able to look after all of their needs but they need a kind of the point of entry and they need a trusted partner to make this happen at a reasonable cost because again, they’re not meta. They’re not Google. They don’t have billions of dollars or even millions of dollars of R&D budgets. It’s like, no, I have a six-person video team. We’re out shooting this candidate or shooting that sports team. We know how much footage we have, often in the hundreds of terabytes, but nobody’s talking millions of dollars to analyze or repurpose this. I’ve been in technology for decades now, and I have never seen a space evolve as quickly as this one, which is both entertaining and challenging but also rewarding because the customers know they need it, and they see all this great stuff they often don’t have the in-house technical teams to do it for themselves.
Advice for AI Founders in the Video Industry
Ryan: I mean, I think that’s a great summary for us to talk about. Again, we talked about where we came from, what’s happening now, where we’re going here. I kind of want, as we’re closing off here, to talk a little bit more, just maybe your advice. For fellow AI founders that are looking to make an impact in the video industry, I’m sure you’ve had people say, Sam, help me out here. What do I do? How do I bring this forward? What would your advice be? What would you offer based on your experiences and what you’ve done with Axle.ai?
Sam: Thanks. The biggest piece of advice I can give is to know the space. Really learn about the workflows of your customers. Make sure that you’re not just producing hammers that don’t fit the actual nails that they have. So, if you know what your customers need and you stay close to that, then you can build solutions that make sense for them. There has been a huge amount of technology-driven development in the space, and there still is right up to this moment where people are just building incredibly cool things because they can and then seeing what sticks. We take a different approach. We know what our customers need. There is something approaching a million video teams worldwide now that do this kind of work, and we know that market really, really well. And we know their workflows and their technologies and their limitations and so we’re able to kind of craft something that makes sense for that audience. So I would say whatever audience that you’re targeting with AI right now, AI itself is commoditizing. There’s so many sources and cool alternatives for any given solution that you’re not going to be able to invent like the only AI engine that does X. But what you can do is be relevant for your customers and solve real-world problems, and that, to me, is the holy grail.
Ryan: That’s amazing, and I think that’s wonderful advice, and again, that customer client-centric and market-centric approach, like making sure you’re listening and growing it that way. And you mentioned there are a million teams. I can’t even imagine how exponentially fast that’s going to grow over the next, as you said, the next, I was going to say a decade, but it’s probably a year, three years, right? In terms of these companies realizing that this is the medium, they need more content. This is the strategy behind doing it. So, Sam, it’s wonderful hearing from you and being able to put this into reality and showcase that to our listeners. Finally, before we go, tell us a bit more about Axle.ai and how our audience can get in contact with you. Learn more about it and just even tap you on the shoulder a little bit if they have some questions about the industry or the space or anything along those lines.
Sam: Absolutely. As you pointed out at the beginning, our website is www.axle.ai. We do hold frequent webinars. We had a very well-attended one at the end of December, and we’re doing another one on February 1 about our new dashboard, AI dashboard technology, where you can oversee a lot of your workflows, but we try to keep that very current. Also, feel free to connect with me on LinkedIn or even email me. My email is sam@axle.ai. All those are good ways to get in touch. Then we have excellent people. We’re quite a child of the COVID years. We’re a very distributed company. We have people in the Asia Pacific, Europe, and Latin America, very talented people who can help you solve these problems wherever you’re located.
Ryan: Fantastic. Sam, thank you so much. Yeah, on the website. Definitely go check it out. I think you can even replay the recent events I see here. I’m browsing it right now of you speaking about the webinar for exploring AI for your media from November. You can go and get caught up on all this great information that’s there for you. It’s at your fingertips for our listeners. Definitely, go take advantage of that. Sam, thank you again so much for your time today and for sharing all of your information with our audiences. You’re just a phenomenal guest, and I’m really excited to dive in a little bit more on this. Hopefully, we can have you back again, too, and talk about some of the other areas of innovation that you’re working on and give some more for our audience here, as I’m sure they’re going to be asking for more.
Sam: I’d be delighted, and it’s been a real pleasure. Thanks so much.
Ryan: Thanks, Sam. I really appreciate it. So with that, I’m going to say to our audience, thank you for joining us on this enlightening journey through AI innovation, and we hope that you’ve been inspired by the incredible stories that have been shared today and remember the future is driven by pioneers like our guests, Sam Bogoch and the limitless possibilities of AI. So stay curious, stay innovative, and keep exploring the boundless horizons of technology. And before we sign off, a small request for our dedicated listeners: as always, if you’ve enjoyed the podcast, I know you have. Please take a moment, leave a review, share it with people that would be interested in listening to Sam as well, and subscribe to the show on your favorite platform. Your feedback, your support, and helping us grow this is what helps us bring more incredible guests like Sam to you each week. So, with that, I would like to sign off. This is Ryan Davies saying thank you again for listening and take care, everybody.