Voices of Video

Why Your Storage Needs a Diet Plan and AI Is the Nutritionist

• NETINT Technologies • Season 3 • Episode 13

Share episode

Meet Predrag Mandelbaum, the innovative force behind MoreScreens - a company on a mission to deliver video to every device imaginable. From smartphones to smart TVs, set-top boxes to game consoles, they're tackling one of streaming's greatest challenges: multi-device compatibility.
 
 Founded in 1999 with live events in Italy, MoreScreens has evolved from basic streaming into a complete end-to-end OTT solution provider. What began as live video delivery gradually expanded to include VOD, content management systems, and multi-device support as customer demands grew. Today, they power streaming services for telecom operators, broadcasters, content providers, and sports organizations across Europe.
 
 The conversation delivers fascinating insights into European streaming infrastructure, where regulatory requirements often dictate on-premises deployments rather than public cloud solutions. GDPR compliance and data sovereignty concerns create a distinctly different operational landscape compared to markets like the United States. Predrag notes that while public cloud services might offer flexibility, many European operators prefer keeping infrastructure in-country for both regulatory compliance and cost control.
 
 Optimization emerges as MoreScreens' defining philosophy. Predrag shares a remarkable customer migration where they reduced server count from 25 to just 6 - a testament to their engineering efficiency. Their pragmatic approach to AI yields equally impressive results: an SVOD provider using their AI-powered personalized messaging saw a 15% reduction in subscriber churn when users attempted to cancel.
 
 The discussion covers the future of streaming technologies, including AI applications for content discovery, metadata enrichment, and storage optimization. Their "AI storage optimizer" leverages machine learning to predict popular content and intelligently cache it, potentially reducing storage requirements tenfold - critical savings for operators managing hundreds of catch-up TV channels.
 
 Whether you're a streaming technology enthusiast, video engineer, or content provider navigating the complex world of multi-device delivery, this conversation provides valuable insights into the technical challenges and innovative solutions shaping the future of video streaming. Check out morescreens.com to learn how they're making "video on every device" a reality.

Stay tuned for more in-depth insights on video technology, trends, and practical applications. Subscribe to Voices of Video: Inside the Tech for exclusive, hands-on knowledge from the experts. For more resources, visit Voices of Video.

Mark Donnigan:

Voices of Video. Voices of Video. Voices of Video.

Predrag Mandlbaum:

Voices of Video.

Mark Donnigan:

Okay, well, good morning, good afternoon, good evening, wherever you are in the world, we're so happy to have you on Voices of Video and again we're bringing you another excellent interview conversation. We like to host real conversations here on Voices of Video, so thank you for joining us. If you're watching on the replay, by the way, you do have the option to watch live. So an advantage to watching live is you can pose questions for our guests, and we always get to our questions, to our live questions, so it's a bit more interactive that way. But let's go ahead and jump in here, and today I am joined by Predrag Mandelbaum. So, first of all, thank you for joining us, predrag. Thank you, mark. Yeah, so you're here to talk to us about more screens, and why don't we just jump in there and tell us what you do? What is Morse Screens?

Predrag Mandlbaum:

What's Morse Screens? Morse Screens is a company specializing in, let's say, end-to-end OTT IPTV solution, services based on our in-house products. Solution and services based on our in-house products. Let's say, our mission is to deliver video and streams on any screen, any device, as we like to say.

Mark Donnigan:

Yeah.

Predrag Mandlbaum:

Yeah, I don't know if you would like to hear more about it.

Mark Donnigan:

Yeah, so it's a big mission. When you talk about delivering video to every device, that's actually very hard to do. So yeah, yeah, exactly exactly. It's hard enough to deliver one stream reliably and in good quality to one device and in good quality to one device. Much less. You know, your. You know all the various mobile phones and car entertainment systems and connected televisions and game consoles, media players, all of that.

Predrag Mandlbaum:

So why don't?

Mark Donnigan:

you? Yeah, yeah, yeah. So maybe you can give us a really quick before we talk about your product and the technology, tell us how you began. What's the history of the company? How did it start? Why did you start it? We?

Predrag Mandlbaum:

simply started with live events in 1999, when we built the first company in Italy, and that was the beginning. So, starting from live streaming at that time using real networks, windows media services, then Flash and so on, and easily, you know, demand was also for video on demand. Then we started producing video, video demand recording streams. Then, let's say, customers asked okay, but how are we going to manage them? So we started building a kind of content management system that we at that time called video management system and that was beginning, let's say, of 40-tier web TV platforms and it is easily going beyond the future management system. And that was beginning, let's say, of 40-tier web TV platforms and it is easily going beyond the future. Then we had to deliver that to mobile, to smart TVs, to different smart TVs and more smart TVs and more set-to-boxes and so on. So let's say that was generally like for last 10 years, interesting. Yeah, that was the journey for the last five years.

Mark Donnigan:

Interesting your engineering team. Where are they based and what does the structure of the company look like?

Predrag Mandlbaum:

We have several offices, let's say the headquarters are in Bosnia-Her Herzegovina. This is the main engineering team. Then we have another two offices in Serbia and Croatia, which are more for support and operations. Plus we have another, let's say the fourth, office which we use for our Italian customers, which is based in Italy. So let's say, engineering gets done centrally in Bosnia, consisting of 30 good, very skilled engineers, and the rest of the team is everywhere.

Mark Donnigan:

Everywhere. Yeah, that's a modern company today. Yes, it is yeah, for sure, for sure. Well, very interesting. Well, why don't you tell us you have built a product and you know, so I'll let you introduce it and then, you know, tell us exactly what it can do and how it works?

Predrag Mandlbaum:

Yes, it's okay, I don't know where to start from. It's simply, you know what it consists of several ecosystem components, such as CDN, cloud, drm, back-office management, subscription management, so on, plus client apps. So what we provide as a software company and SaaS company? We provide, let's say, a full service platform in the cloud which allows customers to manage content, to adjust content, to transcode live VOD audio on demand, to catalog the content and then, of course, to decide the rules for distribution. To decide the rules for distribution, locations, geographies, whatever. Plus, of course, at the top behind, there are analytics, very deep analytics, detailed analytics, reporting, part user management and, in case there is a need for integration, there are open APIs to integrate with any kind of payment processing system and so on.

Mark Donnigan:

So let's say that's in short, yeah, so you know our listeners are typically engineers or you know they're involved in, maybe, engineering management, so they're very technical. Maybe you can share you know. Are you developing all of the components from the ground up? Are you leveraging some commercial technologies and solutions? Are you doing both? You know? Explain to us how this is built and, if you can, you know what solutions you are using. So, if it's open source projects, you know what solutions you are using. So if it's open source projects, you know what are you building on.

Predrag Mandlbaum:

I think that, as any other company, we started from open source, leveraging most of good open source technologies such as NGINX, zabbix for monitoring, elasticsearch, for analytics. Then, of course, there is a component of Python, java, php or more than HTML5 technologies for client apps and native apps if needed. We also built ARM packager for CDN part, so it means like for packaging dash HLS on the flight encryption. The only third party that we are using currently is, of course, widevine or Microsoft Play Ready or Apple Fair Play for DRM services.

Mark Donnigan:

For DRM. Yeah, exactly, interesting. Okay, and are you, do your customers bring the content, or do you bring content to them as well? I mean, is it a complete white label, ott service, content and everything, or do they have to bring their own? No, they have to bring their own content.

Predrag Mandlbaum:

We don't deal with the content, unfortunately. Even we have a lot of requests for such a service of course, of course we are engineering company building technologies, yeah, so it's completely another business for us. But let's say all of them. They have unquoted licensing rights yes, makes sense, makes sense.

Mark Donnigan:

I uh you, through the years, at various points I've been involved in attempting, you know, like you say, to support kind of an end-to-end with content. It's tough, you know it's very tough, unless you are the, you know are, the licensor of that or the licensee of that content. It's hard to sort of two-step content and really build a business. You just can't get the premium stuff.

Predrag Mandlbaum:

That's what everybody wants.

Mark Donnigan:

Yeah, exactly. So let's go back and talk about the devices, because your mission is quite big to support every device. So today, are your customers primarily delivering to mobile? Are your customers primarily delivering to mobile? I notice, you know, set-top boxes are really featured on your website. That's very important. Of course, connected televisions. So maybe you can tell us you know what devices are kind of most popular and what do you find that you really have to focus on supporting?

Predrag Mandlbaum:

Let's say that today, things haven't changed for the last three, four years. So today, mobile it's a must, it's a must.

Mark Donnigan:

Of course, ios Android.

Predrag Mandlbaum:

Yes, that's always Because it's also used to facilitate provisioning on big screens such as Samsung, lg, heizens. In Europe, let's say, in the United States it's mostly around the local or Apple TV, which in Europe we don't have that much, that much, and then Fire TV as well. But let's say, today I would call it mobile, samsung, lg, europe Hisense as well, united States mobile Roku and Fire TV. And, of course, for telecom operators, settobox is still mandatory and it's Android TV as well. I forgot to mention.

Mark Donnigan:

Yeah, are you seeing growth in Android TV? I know that a lot of the Set-top Boxes are running the Android TV OS stack.

Predrag Mandlbaum:

Almost most of them are Android TV or Android ASP still, but in terms of smart TV devices, let's say, I have no recent data. Android TV is growing. Even Samsung and LG are the first two levels and in the US, in the first place, it's a rock. Android TV is something that still has to grow, yeah, but it's very popular. There are, of course, retail Android TV set-to-boxes, which are used by many people around.

Mark Donnigan:

Yeah, I understand. So let's go back to the streaming again. A lot of our listeners, as I mentioned earlier, are technical, they're engineers or they're involved in engineering decisions and they're especially usually in the video encoding area. A lot of the interest the questions we get are around codecs Is AV1 being adopted? What percentage of streams are HEVC? Are you distributing mostly HD? Do you distribute UHD? All those kinds of things Maybe I would like to know.

Mark Donnigan:

There's a lot of talk, at least in the us, and I'd be curious in europe if this is also true. Um, a lot of services are adding a live streaming uh, you know, component to their, to their package, if you want to talk about it that way. And uh, you know, for some people, like you know, like component to their, to their package, if you want to talk about it that way. And uh, you know, for some people, like, uh, you know, like a Netflix is one example. Um, the live streams are reserved to, um, uh, you know somewhat infrequent but very high profile live events, you know, so, like, they recently hosted the Mike Tyson fight with a social media influencer and they've done, I believe, to date, nine events.

Mark Donnigan:

Maybe there's another one that snuck in there that I'm not accounting for. So there's that business model. And then there's all the way to services that have like linear TV. So they still have all the VOD. You can still select a movie or you can binge watch your favorite TV show, but then they have live TV. What's the business model in Europe? Is it similar or is it all VOD, file-based content?

Predrag Mandlbaum:

Is it similar or is it all VOD, file-based content? Look, it's really. You know that Europe is a pretty fragmented continent, so each country has its own preferences. But, let's say, most of telecoms, for instance, they are very traditional, so they are offering Glinar TV, k ketchup tv as well, plus vod, and there are few of them, uh, offering kind of pay tv services like pay-per-view events. Yeah, you know, we really have no reason for that, because it's probably because of, uh, sports rights, because that's the only content that works for pay-per-view. But, let's say, the market is really fragmented but it's mostly linear TV, catch-up plus VOD. Okay, yes, they have tried in the past providing like transactional VOD stuff like that, but people are not using it anymore.

Mark Donnigan:

With Netflix having everything on subscription-based you know it's simply like that, but people are not used anymore, with the netflix having everything on subscription based. You know it's simply like that. Yes, does your customer operate the software? You know, license your software and operate it. Are you running on a public cloud? Talk to me about your architecture.

Predrag Mandlbaum:

Okay, so we have, let's say we have two kinds of deployments, mostly for T2 operators or T1. We are providing on-prem, so it means in their data center all the software components, transcodingn and so on. For the cloud, we are running our cloud, so we list like dozens of data centers worldwide having our cluster and this is, let's say, we call it more screen cloud, but crazy, but it's like that and I simply get access through VPN to the back office and then I can manage the services from there. So it's on the same part, transcoding if needed. Sometimes we do hybrid deployments, so it means like having the storage or transcoding part encoding whatever on-premises customer's premises, and the rest it's in cloud. Okay, so let's look at a solution for that.

Mark Donnigan:

So I notice you have, I think you call it MT Live and you call it a video processing appliance. Yeah, yeah. So what is in that? Is it all CPU running software? Are you using GPU? What are you doing in that box?

Predrag Mandlbaum:

Yes, it was based. Actually we are to replace that unit because current hardware manufacturer announced end of life, unfortunately.

Mark Donnigan:

That's always sad when that happens.

Predrag Mandlbaum:

And that's exactly why we are also considering that, which is great. We are also considering that, which is great. Uh, let's say uh, this product is based on intel quick sync, sdk, and it was really great product at that time because we could save a lot of power, energy to have a very compact two rack units delivering like hundreds of channels in the same unit, plus having the packaging like CDN origin for catch-up TV services on the fly, encryption and everything, and that's something that we are considering now, replacing that one. We are also a software company, but we had to make appliance because that was pretty demand from our customers to have something simple that works and cost effective as well yeah, yeah hey, it's like 18 processors inside, uh, each unit, uh, intel I can't remember even name and uh, yes, uh, interp sync was sdk for that plus to

Mark Donnigan:

manage okay, and and and so so. So you built this um on using ffm peg that's your media processing framework and um and you're using quick sync makes sense. Well, I've got super good news for you um, you're gonna like this. Um, we can cut your rack space in half, go to one ru from two ru and we can give you 50 more hd channels and we can give you av1, hev, h.264. So that's super good news. So you're going to like it.

Predrag Mandlbaum:

That's a great news because optimization is in our DNA. That's why we built most of the components in-house and we really like. We think that you know, iptv OTT, it's always like hard to say they are not so profitable business. The OTT, it's always like hard to say they are not so profitable business. And of course, it's not all about technology. It's also called the rights, which are pretty expensive and so on. But at the end let's say our belief and vision is to optimize as much as possible, anything from technology to the operations.

Mark Donnigan:

Okay, yeah yeah, and when you, when you say optimize um because I I love the word optimize, I I use it a lot myself um, just I find myself using the word optimized a lot professionally in my personal life. But it means different things, right? So sometimes optimized is reducing costs, sometimes optimized is improving performance, sometimes we want it to be both. But maybe you can give some very practical examples of how you're optimizing, because I do find that a lot of vendors, um, everyone uses the words you know and says you know we have an optimized solution?

Mark Donnigan:

Well, but what does that mean?

Predrag Mandlbaum:

So you know this is your chance to say.

Mark Donnigan:

what does it mean? How are you optimizing across your full workflow and all the solutions that you offer?

Predrag Mandlbaum:

Okay, as I said, I also like world optimization. I will give you an example. I won't mention names, but recently we have migrated existing platform to the new one, which is our platform. Okay, they've been running that platform on 25 servers. Okay, we did it on six. Yeah, yeah, that's all optimization, that's that's called optimization.

Predrag Mandlbaum:

Yeah, the same was with empty line, like at that time. They are suddenly like having six, seven, eight CPU-based servers, or even with NVIDIA cards, and they replaced that with the Turek units, or the latest example we are working now on. That's pretty new and we're going to announce the product in 2025. It's, let's say, we call it AI storage optimizer. We show that we, let's say, for customers having like operators running services for more than 100K subscribers, there is a huge overload on the storage for ketchup, so we have to optimize that by 10 times, like replacing 10 storage units with one unit only.

Mark Donnigan:

Yeah, and so how are you doing that?

Predrag Mandlbaum:

Ah, it's a secret, oh, okay, yes, but we invented a kind of new file system and then, thanks to AI, we are predicting the most viewed content or most used content and caching them automatically to the cache engines. So it really reduces overload, because the problem with the storage is bandwidth. At the end it writes and reads at the same time. More customers you have, more users are watching different content. Then it gets stuck. Then you have to add another storage, add another, add another. Imagine operators running like 200 ketchup TV services. It's a huge investment.

Mark Donnigan:

Yeah, operators running like 200 ketchup tv services, it's a huge investment. Yeah, so it must be the case then in in europe that there's not a unique copy requirement in the us, uh, and maybe other parts of the world, but I know in the us, um, there's a lot of content owners that have a unique copy requirement, which means that you can store that content but you and I could be watching the exact same stream or the same program, but my copy has to be unique from yours, so that makes it trickier to do what you're talking about.

Predrag Mandlbaum:

for example, we have it working with Deutsche Telekom for MPVR services. It has to be a unique copy for each subscriber.

Mark Donnigan:

Yeah.

Predrag Mandlbaum:

It's complicated.

Mark Donnigan:

yes, yeah, yeah, yeah.

Predrag Mandlbaum:

It's a shared copy or a unique copy? Yes, yeah, exactly For ketchup. It's completely different, completely different yeah, it's different, yeah, interesting.

Mark Donnigan:

Um well, that I mean that's, that's an impressive optimization, you know, there's um.

Mark Donnigan:

ibc was really interesting. Um, your company was at IBC. Did you show? Yes, Okay, yeah, and I'd be curious to hear if you transition completely off the cloud, off the public cloud, into, you know, private data centers definitely reduce the dependency, you know. So I think kind of hybrid is kind of how most people talk about it. You know where there's private data centers and then, you know, maybe you flex into the cloud for added capacity. This really seems to be a pretty significant trend. Now it's also very difficult for a lot of services because, you know, over the years their platforms have been intricately interwoven, interwoven with the services of some of these large public clouds, and so it's not as simple as just moving your containers to your on-premise servers.

Mark Donnigan:

But let's set that aside. So my point is that what we heard very plainly is a lot of operators said yes, we want to move off, we want to run our own infrastructure, we think we can save money, and there's various reasons to do it. Cost savings is number one, at least that we heard. But there's flexibility, you have control, and I think there's people who are starting to think that way. I'm curious are you seeing the same thing? Are your customers telling you that, or are they going the other way? Are they wanting to go to the public cloud? What are you hearing and seeing?

Predrag Mandlbaum:

Yes, I think there are two reasons why they do this. First, it's especially Europe. Of course, the cost is the first one, but maybe not the most important. But we have really strict user protection regulations. Yes, so, you do not feel so confident sharing databases. There is a GDPR law, stuff like that. So in most cases also from our experience working with public broadcasters, for instance, by law we have to host everything in the country In the country.

Mark Donnigan:

Yeah, that's right.

Predrag Mandlbaum:

Yes, of course we can instance public cloud, but why, what for? Yeah, yeah, yeah. So it's like yes, it's mostly about security stuff, about this user data protection and the cost, of course, because in some cases, all those big guys are sometimes like X5 times more expensive than alternative solutions, alternative cloud providers. Yeah, I think that's one of the most important reasons. It's simply the data protection and somehow maybe lack of trust, you know they share their content somewhere.

Predrag Mandlbaum:

Yeah, I understand. Yes, but most of operators let's say telcos they prefer running everything in their cloud or their data center. That's the rule. For others who doesn't have like skills to manage our cloud, of course they call companies like us asking us if we can build something for them. Manage that, yeah.

Mark Donnigan:

Yeah, yeah, yeah, it makes sense. You know what you're pointing out and for our listeners who are not maybe as familiar with Europe is with the way the market functions and works in Europe, even though I think most services operate in a region, which means multiple countries, but quite often there does have to be almost like standalone systems that are operating in each country. You might be in the Nordics, for example, and of course you understand the breakdown of where this is more a strict requirement, but you know. So you might say, hey, I operate in the region of the Nordics, just as an example. But you know you're going to have oftentimes, you know, sort of dedicated infrastructure within each country that makes up the region called the nordics.

Mark Donnigan:

And you know, and this is just very, very different than the us um, you know where you've got this massive um land mass. You know, and you can distribute freely um anywhere and and even go outside the? U and stream in a lot of places freely. So it does. It creates its own challenges, right, you know. So I'm not saying that that's easier, but it is very different. Do you find, pre-drag, that you are often needing to license the software, often needing to license the software, in other words, sell and license a solution, but also operate the solution? Or are your customers able to take it and say thank you very much, we'll operate it, maintain it.

Predrag Mandlbaum:

Yes, we have different licensing models, but once you know, once you start this kind of business, especially what it is? A kind of a merge. So it's a long term, term relationship. So we have to work closely as one team to operate the platform with them, to help them, to support them and of course, platform has to evolve. So there are often many new things to be added, developed and so on. So let's say in most cases there is no one type relationship like yeah, how to say it? Like we had the key, so that's it. Let's say it's a long term and of course it works also as a pay-as-you-go basis. So more customers they have, they have to license. So let's say we do operate a lot.

Mark Donnigan:

I would, I would assume. So yeah, well, I mentioned at the start of this episode that, for those listening live, you are free to enter questions in the chat. There's a couple of different ways you can get questions to us, but we actually have some pre-registered attendees who already sent their questions in, so why don't we jump into some of these? And I'm taking these in no particular order, so I'm just going to start at the top and work my way down. The first one here is you know, maybe you can comment even more specifically about the changes that you're seeing in the media industry with AI, and I think it'd be good to answer that question from two perspectives.

Mark Donnigan:

So sort of the immediate, like today. How is AI? And you've already mentioned that you were able to get some storage efficiency using AI, so you've given one example, but tell us more like today, and then give us a little bit of the future. You know, where are you building additional capabilities, efficiencies, optimizations since we both like that word using AI, and just give us your perspective there. Okay, using AI and just give us your perspective there.

Predrag Mandlbaum:

Okay, you know all of us, including our customers and us as vendors, we all have different expectations from AI, and that's something that we could discuss for years or days. But to start, let's say, okay, I would divide it into two different operations. Okay, from one side, then you have all this stuff about the marketing, user engagement and so on, and there it is today. So most of them are expecting AI to do work for them, which is not that easy. For instance, if we take today, content owners or public broadcasters or any kind of broadcaster, they are looking to use AI, let's say, for dubbing or subtitling services. That's their work. This is what they see. From other side, we do have, like this, ai optimization of CDNN, like smart caching or smart storing or even smart encoding, and that's one Marketing guys. They would like to have great insights and reports to understand user behavior, to understand the pipeline, to have a kind of predictions and the pipeline to have a kind of predictions, and then support teams as well, or let's say, sales team, to predict channel. That's all doable today.

Predrag Mandlbaum:

But let's say what you are doing today. It's a kind to, let's say, working exactly to reduce the support, let's say, end-user support and to improve or to grow engagement, and we do it in two different directions. One, it's to get a kind of, let's say, tracking or activities, profiling users, segmenting users. We are able to provide recommendation services or, let's say, custom notifications, or to provide them different offers, different packages, everything which can help to upsell the services. And, from other side, for support, guys created a kind of virtual assistant which can help to guide you through the platform, to help you to find the content, to discover the content, to solve the issues. That's something that we are also working on. So there are really many, many things about ai and many different expectations and we will see. Even we also. We are working also on a very interesting feature which we also understood working for all those years with customers Once you provide a report to them of usage of the platform this analyst to explain that.

Predrag Mandlbaum:

So AI, it's a good thing to think about, like creating executive summary of reports, for instance.

Mark Donnigan:

Yeah, that definitely makes sense. I mean, I think we all know that where ai just really shines is, um, is making sense of structured, uh and unstructured data, you know um. So, like you say, being that analyst who now, rather than just handing off a report which literally almost no one reads and those who try to read it, uh can't, you know, just don't have the time to make sense of it and really get to the insights, you know, um, that is an absolute, uh, perfect application for where ai and these large language models you know, when we're saying ai, you know we're really talking about applications for these large language models.

Predrag Mandlbaum:

You know when we're saying AI.

Mark Donnigan:

You know we're really talking about applications for these large language models, these LLMs. I mean, they just absolutely shine, you know there, and so I'm I'm seeing the same thing. I guess I'm curious Again, you know, going going back to IBC, you know, just just just a couple months ago, there was AI on everyone's booth. There was not a single booth that didn't have AI in the signage. And all over AI demo, everybody was doing their AI demo, over um ai demo. You know, everybody was doing their ai demo, um, and yet I think we all, yeah, I think we all you know who kind of went and looked and saw a few of the demos. You know, you know there were some very interesting applications.

Mark Donnigan:

So I'm not saying that they were all sort of uh, fake um, but you know, let's face it, some were um, not as real as as what, as real as what they were claiming. So where do you see the state of the application of AI in terms of some of these use cases that you mentioned? Are they real? Are they in market? Start by talking about your own, your own solutions. You know, like, for example, are you doing um, uh dubbing right now? Uh, via ai? Are you doing closed captioning via ai? Are you doing translation? You know, and those are use cases I would think a lot of your customers would be very, very interested in yes, they, yes, they are definitely, especially, let's say, metadata enrichment.

Mark Donnigan:

Yeah, exactly. Well, that's another one that the large language models are really good at. I mean that's….

Predrag Mandlbaum:

And especially, let's say, we are also working on, let's say, testing different language models and trying to contextualize the video in order to make it more searchable with more context. You know stuff like that. Also for automated cataloging imagine like, usually, an average broadcaster is having from 70 to 150 assets on the platform which you will never discover without a kind of tool, is having from 70 to 150K assets on the platform, yeah, which you will never discover without a kind of tool. So, yes, we are working.

Predrag Mandlbaum:

I was also looking at AI like a buzzword until last year, but finally there are really cool products starting, let's say, on the market which are using AI as a basis and foundation. That's good. So I think, look, it will never replace human, but it will help a lot and it will be our partner for sure. So we have to invest a lot to make different POCs with customers to try to see, and then we will find the right way how to use it. But again, we haven't really need what they are looking for. But I think only the motivation is different. They like to save the money and they like to sell it. So that's the problem.

Mark Donnigan:

Yeah, for sure. I think that's often where innovators and and disruptors, um, sometimes get a little bit confused. Um is that? You know, it's almost like the assumption is, hey, because this is new and cool and innovative. Um, uh, that's value and it's like no. Yes, of course, you know engineers are always technically curious and you know we sell to engineers, so obviously they want to hear about it, they want to see. Of course you know they're going to say, oh, show me what you're doing with AI, but they're not willing to pay just simply because I'm doing some function with AI. It's the benefit that they pay for, it's that optimization, the cost savings, it's the improvement in quality, it's the improvement in monetization capability because now, you know, I have the ability to give, you know, a better personal, personalized experience, for example, you know, and I can serve higher quality ads. You know, advertising, yeah, meaning more relevant, you know, to each viewer. I mean, it's all those things. The fact that it's AI doing it, it kind of doesn't matter, you know, it's the benefit of what it brings.

Predrag Mandlbaum:

Yes, you are completely right, and I will give you another example. We are just at the start, but we also have one customer it's SVOD platform, netflix-like, have one customer it's svod platform, netflix, like you know, and that guy was uh simply providing users, like with the trial package, you know, and uh, you know how users are and they subscribe to turn for seven days, watch everything for free. Then they subscribe to the second, third account to watch again, or they never, uh, or they simply. So what we did? Uh, we created a guide. You know, when you click unsubscribe to get like a kind of personal message uh, hello, mark, but you know we are sorry that you are leaving, but you know you have missed this, this and that you are leaving, but you know you have missed this, this and that you should blah, blah, blah. And then we reduced trend by 15%.

Mark Donnigan:

That's amazing, yeah. And so in that personal message were you looking at like content that I had viewed, so that you were even being very hyper relevant to me. So if I was watching a lot of I don't know uh, you know, action type movies, maybe you would say oh, we're so sorry to see you leave. Uh, you know, um, before you go, maybe you want to watch and you recommend yeah, that's um, yeah, that's really brilliant.

Mark Donnigan:

And, by the way, that answers a question that we have here of how can you leverage AI to increase engagement, and I really like talking about these use cases. Even though we are in the encoding business and we're, in fact, on our ASICs, we can. On our VPU, there's a compute block dedicated to AI so you can do some really cool things right on the silicon. It's really really powerful. So we are all about AI and we're going to be showing some very interesting auto-encoder setup where we're analyzing the incoming stream and then we optimize all the parameters for on the application of machine learning and specifically AI to optimize video. But I really like to also come back to but at the end of the day, it's about business impact and, like the example you just gave, is business impact If you reduce churn by? You said 15%, 15%, yes, yeah, that is. There is not a service provider anywhere that wouldn't say tell me how to do that.

Predrag Mandlbaum:

I hope they tell us to explain that.

Mark Donnigan:

Yeah, yeah, so that's really good, okay. Okay, let's go through some some more questions here. Um, uh, the next question is what's the bottleneck for successful streaming? So I, I, I think that's actually a good question, you know, um, where's the bottleneck today?

Predrag Mandlbaum:

bottlene Bottleneck. It's mostly on CDN part. Last month it's usually last month it's not even CDN, you know, sometimes. Let's take, for instance, the last live event of Netflix, which is great company with a huge network. Of course they are specializing more in BOD delivery. Yeah, for live. I really don't know any data on how they do what they do, but let's take an example how many millions they had during that lifetime? I think most of the problems have been on the last mile, because sometimes you can really you can't control it. Yeah, now, close collaboration between CDNs and, let's say, tier 1, tier 2 operators, but even tier 3. Yeah, stuff like that.

Mark Donnigan:

Yeah. So it's an interesting observation. And again I'm wondering if you can comment on the difference between how CDNs need to operate in Europe versus kind of the rest of the world, where in Europe again getting back to the self-contained you know there's sometimes the need. Not sometimes, quite often, because of GDPR there is the need, especially if you're doing personalized services you have to use a CDN in that country and then the next country right next door, you have to use a CDN in that. So people talk about multi-CDN strategies.

Predrag Mandlbaum:

In.

Mark Donnigan:

Europe you have to be multi-CDN, right. Maybe you can give those who aren't as familiar a little bit of a lesson as to how that works in Europe, how CDNs need to operate.

Predrag Mandlbaum:

In Europe there are not that big issues, let's say. But I will give you another example. We just did a project from onei company who has to deliver the content to us, okay, and to distribute channels to us, and first mandatory request was you cannot use a european cdr, everything has to go directly from the source to the US. So we have to build additional nodes inside the US, in Europe, for instance, we see also many operators are even using US servers to deliver content into Europe. So there is no big difference About the content. I see no reason until it works because Europe is really connected. There are huge exchanges in Amsterdam, frankfurt, paris, for instance. So there is no big deal about it. Let's say it simply works. Today Maybe the user is not that big as in the United States, but again, it works.

Mark Donnigan:

Yeah, okay.

Predrag Mandlbaum:

Because most of, let's say, most successful platforms are managed by telecom operators.

Mark Donnigan:

Yeah, yeah, and your customers are mostly operators. Is that your customer segment?

Predrag Mandlbaum:

Yeah, we do have like four lines. It's operators, broadcasters, content providers willing to launch SVOD services and sports. That's the core. In the past we've been working for governments, enterprises, universities when we started. But let's say, those four verticals are the most important ones to us. What we see as a trend more and more publishers are getting into broadcast business, launching new linear channels, which is also active. But let's say, are getting into broadcast business, launching new linear channels, which is also productive, you know. But let's say, 60% of our business is telecom, 25% broadcasters and the rest is yes.

Mark Donnigan:

Yeah, I understand, but sports is very important. Yeah, yeah, interesting, yeah, interesting, okay, well, this has been a a good discussion. Very interesting to hear your perspectives from the. You know from where you sit in the business, in the ecosystem. It's one of the things I love about hosting this show, voices of Video is I get to hear from so many different types of companies, different parts of the world, different technologies. I really get a diversity about what's happening in video. I enjoy that and I hope that our listeners, you know, also appreciate that. We know they do, because they continue to listen. So that's good, thank you, yeah, yeah, well, that's good, well, well, let's, let's wrap it up here. Pre-drag, I really appreciate you coming on and thank you for your time. Um, if people want to learn you know a little bit more about your company, uh, maybe you can share your, your website and, uh, any other details yes, it's morescreenscom very simple yeah, yeah, that's great, okay, well, good, um.

Mark Donnigan:

well, thank you for everyone who joined live we really appreciate that and for those who, uh, you know, who listened on the replay, as we say, uh, remember, you can join live and, uh, you get a chance to answer questions or ask questions, and we'll do our very best to answer them. So, thank you again and, pre-drag, have a great rest of the day.

Predrag Mandlbaum:

Thank you, you too, mark, looking forward to seeing you soon, summer Likewise.

Mark Donnigan:

That's right, that's right.

Predrag Mandlbaum:

This episode of Voices of Video is brought to you by NetInt Technologies. If you are looking for cutting-edge video encoding solutions, check out NetInt's products at netintcom.

People on this episode

Head Shot

Mark Donnigan

Host
Head Shot

Jan Ozer

Co-host
Head Shot

Anita Fejter

Producer
Contributor

Predrag Mandlbaum

Guest