The 16:9 PODCAST IS SPONSORED BY SCREENFEED – DIGITAL SIGNAGE CONTENT
What if you could use AI to make digital signage screen content relentlessly relevant?
That's the premise and promise of what Korbyt calls Machine Learning Broadcast, new capabilities in the Dallas-based software firm's CMS platform.
Using computer vision and machine learning, the idea is that if the platform can get a sense of what's making people stop and watch in a defined environment, then content can be optimized based on that interest.
The system finds and schedules content to push to screens based on engagement metrics.
How it all technically works is a bit over my shiny head, but I had a good chat with Korbyt CTO George Clopp about what's going on and its implications. We also get into what the future looks like for AI in digital signage.
Subscribe from wherever you pick up new podcasts.
TRANSCRIPT
Geroge, thank you for joining me. We've chatted in the past. For those who don't know Korbyt, can you give me a rundown of what the company's all about?
George Clopp: Hi, Dave. It's a pleasure to speak with you again. Yeah, Korbyt is at its root an employee engagement company. So we've got roots in digital signage, but our typical use case is using digital signage at corporate campuses and to communicate to employees, to increase employee engagement as well as to communicate real-time mission-critical stats as well.
Is that pretty much the core vertical that you guys chase, workplace?
George Clopp: It is. We are heavily into the workplace, meeting rooms as well. We do a lot with retail banks, a little bit into the retail space, but it's primarily corporate campuses.
For those who don't know the company, it actually goes back a long way to Symon Communications days, right? You guys were doing workplace communications long before the digital signage industry discovered that.
George Clopp: Yeah, exactly right, Dave. It precedes me. I've been here for seven years now. I can't even believe it, but that's how much I enjoy this space and the industry. I enjoy the company so much, but we had Target Vision, Symon Communications, and we've just evolved. I joined at the tail end of 2016 to develop the Korbyt platform, and obviously, we have to meet the needs of the digital signage industry, but we've had a really heavy focus on employee engagement as well.
Is it interesting to see all these other companies who have more general offers, find their way into the workplace because they see that as an opportune vertical?
George Clopp: Yeah, I view it as exciting. I think it's definitely a macroeconomic trend with the pandemic, post-pandemic, the modern workplace, everything is reimagining and reinventing and re-everything these days.
I think it's good. It's a legitimate macro problem that everyone's looking to provide solutions to. So, I'm really excited. I love the industry myself.
In some respects, you guys have been doing back-of-house, a lot longer than most companies would have. I mean, you're not just working in the offices, you're working in production areas and so on.
George Clopp: That's correct. Heavy in manufacturing and heavy in the contact centers, anytime where you're doing mission-critical real-time data, you're connecting to an ERP (Enterprise Resource Planning), or yard management system, and you want to change or orchestrate the display and the surroundings based on data changing, we've got a deep background in that.
Yeah, for contact centers, if I recall, years ago pre-arrival with the company, you were doing low-resolution LED readouts that were just telling people in the contact center about the average wait time on calls and things like that.
George Clopp: Exactly, and that's matured over the years and now we're doing that on the desktop and on the mobile device as well. We still have some supply chains and some yard management systems in a warehouse, where we'll do the little blinky boards over the dock doors themselves.
We range from the dock doors all the way to your mobile device now.
The PR that came out about a new piece of functionality, your marketing talks about a million endpoints, 250 cloud migrations, and 100+ native integrations.
A million endpoints, that's like a lot.
George Clopp: It is. Yeah, scalability and being able to expand out to touch desktops, normal, typical digital science screens, and mobile endpoints. It's been a real focus on us for the last four or five years. So we're really proud to announce that, and then the back end, like you were talking about those native data integrations, I think that's really what sets us aside from a lot of our competition is making those really hardcore authentications and then that real-time pipe between us and the source systems.
I know a lot of other software in our space that we run into, they talk about integrations. A lot of times it's really just a file, they're taking data from a source system. They're putting it into a CSV format or any kind of other format and then they're pulling that in. So that's really where we shine with that real-time data integration.
Is that important in terms of a distinction when solutions providers and users are looking at data integration and they see that a CMS says, yeah we do data integration, we can integrate with your platform? It sounds like you're saying there are different tiers of that, and there's real integration and there's just like a baseline.
George Clopp: Yeah, exactly. That's the right way to pick up on that day, for sure. When you need to orchestrate and change things in a 911 center or in a manufacturing-type environment and definitely in a contact center, speed is really the key there. So having something on a five-minute loop that's pulling a file, it's just not fast enough. So you need that real-time data, you need that high availability so that something was to break that you've got a backup in place and you can make sure that contact center, that supply chain, that 911 center is rolling smoothly.
They're not just getting their data, but they're changing the experience of the data. That's another thing that we do, we pull in stats, but we also augment those stats and do value-added calculations on the stats, and then we trigger on those values to change the screen, or change the mobile device or change the desktop. So if you've got too many calls in the queue or you're running behind on this loading dock here, we'll change the entire experience for you based on that value-added stat that we do.
I also assume that when companies talk about integrations, for very logical reasons, they're going to go to the most used platforms out there, whether it's Teams or God knows what. But if you have a hundred plus native integrations you're probably talking about some pretty exotic things that nobody's ever heard of, and if a company went in and said, we can integrate with their systems and they say, what those systems are, their eyebrows are going up, because they're thinking, I have never heard of that.
George Clopp: Absolutely, Dave. There are some low-level protocols where we just integrate at a TCP level with a very proprietary protocol, but I would say the bulk of it is more modern, JSON-based RESTful interfaces, for sure and we like to distinguish between data integrations, business application integrations, and SSO integrations, in three categories there.
So, like a Power BI or a Tableau or something like that would be more of a business application integration, and when we're talking data integration, we're talking more low level, running SQL against a data store, running web services, running SOAP-based web services, and to that extent. And again, that's why we call it out in our marketing because we do think that's a core differentiator for us.
So just to go back to something, when you talk about a million endpoints, you're including desktops..
George Clopp: That's correct. Desktops and mobile devices, basically all of the endpoints that we talk to.
Good. Back at the start of summer, you guys introduced something called, Machine Learning Broadcast. What is that?
George Clopp: Yeah, fantastic question. We were involved with machine learning, and AI before it was really cool, so this was actually something we developed in 2018. We've been honing the model, and then we re-released it this year. But machine learning is a subset of AI, and we all know AI is a super big buzzword these days and when you peel that onion, there's levels of accuracy involved there, and there's a lot of hype around the world.
But the reason why we called the feature machine learning broadcast is really to focus on the ML aspects of it, and it's a great business problem to solve because, at the end of the day, what we're really creating is a recommendation engine. And I think everybody's familiar with the Amazon recommendation engine, Instagram, and other social media platforms that are just, they're recommending content for you.
That's essentially what we're doing here. We're using KNN Analysis, which is supervised machine learning to look at content that has some engagement with it, and that engagement could be measured by computer vision on a digital signage screen, it could be measured by interactivity with it on a desktop or interactivity with that content on the mobile device and then behind the scenes, all we're doing is we're finding out second, third, fourth-degree order content, that's related to the content that was engaging and then it's a feedback loop. We go ahead and automatically schedule that content and see how that content is engaged with so it's a self-learning feedback loop there and the whole purpose of it is to find content that's engaging and show more of that content to your employees.
Could you give me a real-world kind of example of how that might work?
George Clopp: Yeah, absolutely, Dave. Let's say a company's opening up a brand new office in Buenos Aires and for whatever reason, people really gravitate to that content. They look at it on the signage screen, on the fifth-floor break room, they're engaging with it on their desktop, they're looking at it on the mobile device. We learn from that engagement and say, okay, let's go ahead and find similar related content there. Let's find content related to office openings in Buenos Aires, and then let's go ahead and go further out and look at second, third-order tags. So that would be content related to South America as well. And then we automatically play that content, inject it back into the playlist, and our customers have complete control over whether it's automatic and which players actually get this content and which devices get it and then, we learn based on that content. So it's a feedback loop, and you might find in that case that your employees are really more interested in the geographic region than they are in the new office opening.
So it's relentlessly relevant.
George Clopp: Exactly right, Dave, and solving a real-world business problem because one of the challenges our customers have is, it's really arduous to constantly schedule new relevant content.
The first couple of times you do it, you create a scheduled playlist. Yeah, it's okay, but it takes a long time and then, with Attention Deficit Disorder in today's modern world, people grow immune, and they tune out that same content over and over again. So, you need that fresh content injected to keep the employee's attention.
I'm guessing that somebody's going to be listening to this and thinking, that's cool, but where on earth do I get, or how do I develop all this content so that I do have this somewhat bottomless hyper-relevant content available?
George Clopp: Yeah, fantastic question. Right now, in its current stance with our ML broadcast, you need to have that content in your media library. We're not automatically going out to like copyright-free areas and pulling in content. But with our release coming out next year, it's called our AI employee engagement. With that, we'll automatically be creating and sourcing content for you on your behalf.
Yeah, I saw a demo of something like that over in Germany a little while back with another company who, I'm sure you'll be happy if I don't name them, that was all about using what was available through an intranet and an extranet, and other resources to auto-generate content for screens.
George Clopp: Yeah, it's opening up the whole world of generative AI. We're actually looking at both. Whether there are generative images, generative video, or generative text. Obviously, in our space, images and videos mean a lot, and there are different systems out there. There's DALI 2, there's stable diffusion. They've all got their strengths and their weaknesses. But we're combining that with templated-based content as well.
So automatically generating content that's relevant based off of a text prompt is super useful. But in some cases, it might not be the right content that's generated. So we also will have a mixture of templated content as well.
Yeah, I think templates are a big part of that. I've farted around with things like Mid Journey and so on, and you could see how it could go sideways on you really quickly if you left too much up to the machine.
George Clopp: Exactly. It gets into that whole thing of prompt engineering.
You got to be really good with your prompts, and they've all got issues like generating hands and things of that nature right now. But we want to be on the leading edge of this, use it where it makes sense. An area where we think it really makes a lot of sense, a preview into our AI Employee Engagement, is on mission values and goals. We feel like that's an area where our customers just don't communicate enough to their employees, like, there's cake in the break room, let's recognize employees.
That's all part of it, but really just reinforcing, Hey, your goal in the finance department this week is to close your books three days earlier. And so, mix that text in with some great video or some great images that are created in the background using this generative AI.
Yeah, I saw something on LinkedIn last night, and I commented on it because I thought it is great that there's a company that's using KPIs and messaging right on the production floor, and the person who posted about it said, this is not very sexy, but it goes to what's needed on the floor for those workers. But the problem was, it looked like hell.
It was just black and white, and they were slapping up a whole bunch of Excel charts, like a stock of them and you'd need binoculars to even see them. So it's important to think about the presentation.
George Clopp: Yeah, totally agree, Dave. I say this at all my speaking events: content is king, content is queen, and that still rules the day.
When we're intermixing real-time data with content, it has to be visually appealing. You can't have 20 different stats on the screen; all of those rules of graphic design, I still think, hold true here.
Do you see a day when things like scheduling and trafficking of content are largely automated and handed off to machine learning or some variant of AI?
George Clopp: That's exactly what we're trying to build, Dave, with a release next year. With the ability, of course, to intervene, the ability for the communicator to come in and approve the content or really go ahead and bias the content and say, okay, I've got these 30 categories of content I see that I really want to bias, what the content areas could be.
“Hey, I'm a new enroll. I'm a new first-time line manager. I'm a new director. I'm a new VP, and there's content associated with that new enroll.” They might want to bias that and increase the weight on it, decrease the weight on it, or take it out altogether. So there's still going to be that human touch involved in the ability to approve content, but the AI itself will take care of making sure that content is fresh and relevant.
And the big problem we're solving there is just that, again, attention deficit disorder people have, if they see the same thing on the screen, week after week, they tend to tune out. So how can we think of innovative ways to display KPIs, display goals, display things that are really important to the company and give it a great background, give it a great video so that it gets employees' attention again?
We're going to talk about machine learning. You reference AI-driven camera optics. Is that basically a computer vision?
George Clopp: It is. Absolutely is, yes.
Did you guys write your own, or are you using something like Intel's OpenVINO?
George Clopp: Yeah, the two big ones out there, we've used OpenCV, that is, Open Computer Vision, and TensorFlow, and they both have their strengths and weaknesses, but there are higher order problems we're trying to solve here, and not reinvent computer vision so we're using some libraries for that.
Is that just part of the mix of doing this sort of thing? Are there other technologies you can use to get a sense of dynamics in a venue?
George Clopp: Yeah, I think so. Infrared detectors, pressure sensors that kind of tell you who's in that immediate vicinity. You're basically correlating that to human beings in the vicinity, how many human beings are there, and what was playing on the screen at that time. Yeah, so there are less technological ways to do this and still get some good results.
AI is being talked about a lot as you've gone through about its potential to automate presentations. Are there other aspects to a digital signage company, the way your company operates, that you can use AI to help with marketing, help with customer contact, that sort of thing?
George Clopp: Yeah, without a doubt. I'm sure you're reading everything. It's revolutionizing all traditional roles, right? Not just engineers writing code. You got a chat with a ChatGPT engineer. With Microsoft's Copilot, it's going to revolutionize the way we all use Excel and Word and PowerPoint and things of that nature.
It's definitely revolutionizing marketing. Building product brochures for you automatically, things of that nature, and then, that naturally progresses into, is AI going to take all of our jobs, which I don't think so, going to help us all become more productive. The employees that really change and adopt the AI, I think they're going to be even more valuable than they are today.
It's just the employees that just say, I'm not going to do this, and they refuse to allow their cheese to be moved, those are the ones that I think you have to watch out for.
There’s an increasing number of companies. I just wrote about one today that has gone down the path of headless CMS. The idea that you can leave the final presentation later, the interactive element, whatever it is to software developers at a large company or who works with a large company as a services company and the digital signage CMS is just the infrastructure, the foundational platform that does device management, scheduling, trafficking, all that sort of stuff.
Are you seeing that demand in the marketplace?
George Clopp: We're seeing the opposite. What you're saying absolutely makes sense, especially with my background and the way we've architected our product with microservices. What we're seeing, especially with our large enterprise customers is, they want a little more white glove service.
Taking on the arduous task of piecing everything together, even with a microservices framework, is putting a lot of ownership on them. But that is not to say that there's not a need out there. We just really haven't found it. We've actually gone the opposite direction on our side, which has really served us well because we've gone from zero revenue in the cloud to 2 million. We brought on a new CEO, and we quickly ramped up to 20 million. I think it's working for us so far.
Yeah, you're a very different company than maybe prior to you joining RMG Networks, that was a weird little side trip into digital out of home.
George Clopp: It was. We see the artifacts and all that, but I think it's a great group of people here now. There's not a leftover where people have bad attitudes or anything like that. So really proud of where the company's been, the talent we've acquired. We've acquired people from all over the industry. Really love working with the current team and cross-functionally, not just engineering and support, which is what I run, but in sales and marketing as well.
Yeah, it's interesting when you mentioned you've gone in the opposite direction of headless. I've heard that as well, particularly when you get into, like Fortune 500, Fortune 100 kinds of enterprise-grade customers. They want to outsource digital signage, by and large, in the same way that they've outsourced a lot of IT services.
George Clopp: Yeah, absolutely. That's the same trend we're seeing, Dave too. It's a little bit of both, right? Everybody wants their cake and eats it too, right? Like they want you to have the ability to do it, but then when it comes time to actually execute on it, we typically find, Hey, we can help them get faster to market if we help augment their team.
How important is security?
George Clopp: Oh! It’s Huge. We all know that the disaster scenario in digital signage, someone compromises your network and they put up some content images or videos that are not appropriate. Even more so with us being more omni-channel with desktop, mobile devices. We've got a data privacy officer, we're SOC 2 compliant. We do a lot of work in Europe so GDPR comes up a lot as well, data privacy. So I think it’s super important.
When I think you look at the different offerings out there and the first tier, we look and sound the same. So I think what you got to do with new prospects or new customers, they just got to peel that onion more. What does that really mean? What does it mean that you encrypt your data? Do you do it at rest? Do you do it in transit? Those kinds of things, and I think that's where you can tell the difference between different offerings.
And are the people in the first and second meetings with prospective customers different than they were 7 years ago when you started? I'm hearing the IT people who used to come to meetings and sit there with their arms crossed, thinking, dear God, how long is this going to go on? They're now tending to lead these meetings.
George Clopp: Yeah, I've seen it in multiple ways. Definitely, IT is still the big persona of the buyer here. But I'm also seeing less and less about speeds and feeds and players and hardware and transmission equipment and scalers and more about the final purpose of what we're trying to do.
I'm just starting to see that shift. Seven years ago, I talked to people, and it's the AV integration guy. I don't really care what's on the screen. I just care that it's not dark. I don't want a screen that's down. That's their most important thing, and now I'm seeing that shift a little bit more towards they do care about the content, and they're bringing in more of the HR and the communications group involved and making sure that the platform can grow. I can create content on the platform or I can integrate with Adobe or SharePoint or something along those lines. But I still see it, especially AV/IT as a huge influence in the buying process.
Yeah, certainly going back seven, eight years when I was doing some one-to-one consulting with enterprise level customers, that sort of thing, I would go into a first meeting, and I would say, okay, why do you want to do this? And it was always intriguing to see how often people would lean back in their chairs and say, I hadn't really thought about that. They wanted this thing, but as you say, they didn't really know what they were going to do with this thing.
George Clopp: Yeah, exactly. And there's a little bit of power in that too. There's power to putting the latest and greatest screen technology in your office and giving you that modern technology look and feel but then just carry it one more step in the maturity direction and start focusing on the content too.
Yeah, you can demonstrate innovation by having a big ass screen in your lobby, but if there's nothing useful on there, you're not really demonstrating a lot of innovation.
George Clopp: Exactly, and I think there's still room for that super wonderful creative experience that's human-curated that graphic designers make, and they spend a lot of time getting just perfect in those high profile areas, like the lobby of a company, and then there's also opportunity for, new content generation automatically for me so that I don't have to necessarily sit here and handle this thing. So I think we're going to live in a world where both will be applicable.
So you mentioned you, you're working on new iterations of AI-driven content. Is that the big kind of roadmap item for your company over the next year?
George Clopp: Yes, it really is. Yeah. We've got a huge, large-player ecosystem, all the data integrations, and omni-channel platforms. So where our new development team is focused on is automating the content creation, automating that entire feed, if you will, so that it really takes that arduous process away from our communicator.
How many folks do you have in the company now?
George Clopp: We're a little under 70 people right now. So still a small company and I love it cause everybody has to wear multiple hats, do multiple roles. You have to bring a lot of energy to the company, and I just love that. I've just grown so fond of it over the last seven years.
And is most of the team in the Dallas Fort Worth area, or are you all over the place?
George Clopp: Since COVID, we're mainly in Dallas, but since COVID, a lot of us have moved out a little bit. So I'm actually in Colorado. Some of my engineering leads are in the West Coast, some are in Pennsylvania. So we're really practicing what we preach, the hybrid workforce.
All right, George, thank you for spending some time with me. It was good to catch up.
George Clopp: Yeah, it's fantastic, Dave. Thank you so much for taking time out.
Pierre Gendron, Stingray Media
Kyle Pilot, iGotcha
David Anderson, Mimo Monitors
Gordon Feller, Meeting Of The Minds (Smart Cities)
Nick Fearnley, Signstix
Art Stavenka, Kino-Mo
Greg Topel, Tanvas
Brett Jones, Lightform
Jonathan Gudai, Adomni
Arie Stavchansky, Dataclay
Mitch Goss, Zero-In
Sheldon Silverman, Smartbomb Media
Matt Gibbs, UPshow
Alan Brawn, Brawn Consulting
Jason Bier, Federation For Internet Alerts
Mark Stross, ANC
Megan Dutta, Women Of Digital Signage
Jenn Vail, E Ink
Reece Kurtenbach, Daktronics
Stan Richter, SignageOS
Create your
podcast in
minutes
It is Free
Insight Story: Tech Trends Unpacked
Zero-Shot
Fast Forward by Tomorrow Unlocked: Tech past, tech future
Lex Fridman Podcast
The Unbelivable Truth - Series 1 - 26 including specials and pilot