Previously featured on Intel® Intelligent Edge video series.

‍

Measure how people work, shop, learn, and play in physical spaces without the need for complex development, with meldCX & Intel. In conversation with Bill Pearson, Intel VP of Internet of Things Group and GM of Developer Enabling.

Featuring:

• Bill Pearson - VP of Internet of Things Group and GM of Developer Enabling at Intel
• Thor Turrecha - EVP of Global SaaS at meldCX

Transcript:

[00:00:00] Bill Pearson: Welcome to the Intel Edge Innovation executive fireside chat. This video series showcases the great things our ecosystem partners are doing with their Intel-based AI enabled offerings. And today we'll be discussing meldCX vision analytics solution Viana with Thor Turrecha. meldCX is an information technology company that empowers businesses and individuals to create premier customer experiences by using AI technologies.

[00:00:39] So welcome Thor. Can you tell us a little bit more about your role and your company solutions?

[00:00:44] Thor Turrecha: Thank you for having me today, Bill. My name is Thor Turrecha and I'm the EVP of Global SAAS here at meldCX. Um, a little bit about meldCX, we focus on melding customer experience. We specialize also in solving complex problems and we use AI and data science to help our customers make better decisions.

[00:01:06] So essentially what we do is we teach cameras, new tricks and to be able to see, analyze and make actionable decisions seamlessly. So with the help of OpenVINO toolkit, we've developed our product called Viana to help organizations harness the power of machine learning and gather insights, make smart decisions and help them automate their process.

[00:01:31] Bill Pearson: Yeah, it seems like something that the world could benefit from. I mean, with the global market for AI and big data and IOT growing rapidly, reaching 26 billion by 2025. Uh, businesses looking for new ways to get more value out of that IOT data, whether it's efficiency or effectiveness or both. Uh, and then this seemingly unending appetite for solutions that can perform, you know, AI tasks on edge devices while consolidating some of the big data processing tasks before they go to the cloud.

[00:02:00] Uh, so companies are seeking new ways of, you know, driving security and, you know, faster bandwidth reduced costs, uh, better latency and just, just increased performance. As this happens, what trends and pain points do you see shaping the current and future state of the industry?

[00:02:16] Thor Turrecha: That's, that's a good point.

[00:02:17] I think the world has changed in the last two years and the rate of change is much more faster. How humans interact with each other is now different. And based on our research and, and my experience, the predominant drivers of the changer are mostly one, the convenience. 

[00:02:40] And convenience means today you let your car drive you around. Um, you let your phone, you talk to your phone and order your food tonight, right. In physical stores you can go in and shop without even interacting to any humans. You can even, you know, do it without actually taking your wallet out. And that, those are amazing kind of experience that is being provided. But one thing's for sure, convenience is becoming more of, you know, kind of an expectation by customers. 

[00:03:20] And number two is experience. Today people interact with technology and, and they're expecting the technology to have a contextual interaction with humans, right? Like a good example is the smart speakers, you know, you are expecting it to recognize your voice, your accent, um, your favorite online store.

[00:03:43] You're expecting it to know the preference, your preference, and recommend new products based on your interest. How you watch your movie, you know, streaming services now are based basically smart enough to recommend what's the next movie you should watch. Your social media account knows you better than your closest friend.

[00:04:04] All these trends continue to evolve, and we will see a higher demand, I think, on edge computing, demand for faster bandwidth, tighter security, and definitely better and elegant solutions as people will likely to expect that online and offline experiences will become seamless.

[00:04:26] Bill Pearson: Yeah, for sure.

[00:04:27] And as the industry is evolving and these expectations are changing and growing, how does your solution help capitalize on these opportunities and solve some of these problems that are coming up?

[00:04:38] Thor Turrecha: As I've mentioned earlier, as humans expect to get the same experience and convenience that is offered online in a physical space, we believe that there is an emerging demand to attract the physical space, like a website.

[00:04:52] We have designed and built Viana to take a sensor and extend its use beyond security and surveillance. Using the camera as a sensor powered by an edge device, we start offering the capability to track how humans interact with each other and to the world. And of course we're doing it in a safe and ethical way.

[00:05:14] Also, Bill, we feel like we're lucky that even despite COVID, we managed to roll our product into three different continents and we've got customers in different verticals - from retail banking, government, food manufacturing. And that's what really resonated with our customers, is that the way we approach AI.

[00:05:35] And what I can think of is that one we really seek to democratize AI. We want to break it down and make it easy. And number two, we take, we're taking a hard kind of flow in terms of, how can we continue to develop AI? And AI is very powerful, but still be ethical. So, you know, having that sort of balance between, making it simple and then at the same time as well, with the power and reach that an AI could provide us, we remain focused on making sure that we are ethical. We do not, like a good example is we do not use random images. From website, we basically create synthetic data, fake data, to be able to create our own and therefore, in a way it's safer and we're not using any other people's content as we continue to do it. It is certainly a bit difficult. It's not as easy as the traditional way of doing it. But those are the main reasons why I think our customers are trusting us. 

[00:06:51] And as I'm diving deep into that, I've got a couple of slides here that I want to show around how we’re actually tracking persona in an ethical way. So on this slide, you can see that I have, on the left side, Joy. And on the right side, you've seen joy space being blurred and that's the sort of, kind of, um, standard process that we do to make sure that we don't disclose any personal identifiable information. And you can see as well that we've tokenized Joy. So instead of calling her “Joy” in our database, she's basically has a token of A392KJ-002.

[00:07:37] So you can see, we can profile Joy using that ID. And as Joy goes through a store, for instance, that has hundreds of cameras. Every time Joy goes into that zone, we can identify her, we know what her behavior is, and we know what kind of product that she's interacting with. And those are rich behavioral information that it's useful for the retailer, but then we're not actually taking Joy’s personal information. 

[00:08:06] At the end of the journey, the customer would just know that there's this person with this classifier as “female young adult”, and she's happy predominantly along the way of experiencing the retail store, and have basically interacted in a couple of products along the way.

[00:08:24] So that information is quite useful in helping retailers, you know, currate, better experience for these types of persona. And I think those are the key things that we've shifted our focus from just focusing on identifying who the person is, but actually not focused on the “who”, but focus on what are the things that they are interested in, what aare the intent, and what are the other behavioral traits. They’ll probably not allow us to capture name, emails, and mobile phones, like the traditional way of tracking marketing information. 

[00:09:01] So I just want to share that, Bil, because I think that's quite important on how we can differentiate ourselves that we have a powerful platform and yet we didn't use that power to basically invade into somebody's privacy and we keep, make sure that we've got a fine line that we follow in terms of being ethical and safe.

[00:09:20] Bill Pearson: Yeah, it seems like you've really thought this through. And, you know, building this powerful capability on this platform of trust could be certainly a differentiator in the market. Are there other things that set you apart in the market? 

[00:09:34] Thor Turrecha: With partners, we certainly see that Viana is easy to sell.

[00:09:39] We have a program where within the sales conversation, if the customer decided “let's run a proof of concept today”, we can make it go live within the day. Also whilst we can start small and quick, we have the ability to actually take any existing models that the customers might have. And with, you know, OpenVINO, we help them optimize that and convert that into OpenVINO and the purpose of converting that to OpenVINO, because it can actually generate more inference based on the existing compute. So it can just do more with less. 

[00:10:20] And then lastly, we do a lot of moonshot projects where customers have personalization requests where their business is quite different and they want to create a feature that's only relevant for them. So we cater for those kinds of services. So that is mostly our key differentiator: that we can go out and sell and it's easy to sell through our partners. We can accommodate new models that are coming from our customers, or anyone basically, they could bring their own model.

[00:10:53] And then lastly that we can personalize our models to create a feature that is more fit to the business. And also it's quite simple for the users. We take pride that we are taking something complex and we made it simple and most of our customers are actually marketers and non-technical individuals.

[00:11:16] Also we are now working within Intel where the project is all about: how do we bring the capability to a student that is a 10 to 15 years old, and they can then train and build their own model and they can deploy it through our platform.

[00:11:33] Bill Pearson: That's amazing. I mean, being customer focused and making things simple as a powerful combination.

[00:11:38] And we were talking about OpenVINO and working with Intel. Can you maybe say more about how working with Intel has been an enabler for you and helped you to solve some of these challenges and opportunities that are in front of you?

[00:11:50] Thor Turrecha: In different kinds of levels, you guys have been really helpful. First from an engineering standpoint, you guys have been really helpful in getting us exposed on most of the engineering activities you guys are doing.

[00:12:06] We collaborate together and we solve epic problems together. We also have experience with Intel that you guys are expanding the sort of influence of, um, kind of the horizon on where we play and where we innovate. One example is that Intel has a lot of universities that are innovating and researching.

[00:12:31] And occasionally we have problems that basically overlaps between the private sector, the education sector, the government, and you bring all those talents together and solve those problems. And I think that helped us sharpen our tools and become better engineers. 

[00:12:50] Also the toolkits that you provide. OpenVINO have really helped us go to market relatively quickly. Viana started pretty much when COVID started and we were able to build the first version of the platform within three, four months - around that time. It's simply because we tap into the ecosystem that OpenVINO’s already provided. The OpenVINO zoo is already there, so we could access to a lot of OpenVINO models that's available. The inferencing engine, and the optimization that allows us to tap even use old CPUs that allows us to infer.

[00:13:33] So yeah, we’re very thankful for the relationship. And then you guys have brought us into problem statements that otherwise we won't have access. And through that, that can have more enjoyable for us and bringing products in the market because you’re, you know, closer to the people that are here.

[00:13:53] Bill Pearson: Fantastic. That's great to hear. Can you maybe talk about a success story, something that stands out for you andm you know, one of your customers in some of the great work that you guys have done together?

[00:14:04] Thor Turrecha: What comes to mind is Westpac. Westpac is one of the big four banks here in Australia. And they have about 800 plus branches.

[00:14:14] And out of those branches, they've got thousands of screens deployed. These are digital signage screens. And one of the problem statement that we're trying to help them solve is that they produce hundreds and hundreds of content. And they don't know which content works. So, in the past, they'd basically just take content that they have been doing on website and just putting it on signage screens. 

[00:14:43] And to that point, they're really struggling to attribute value, as to what value it basically brings in. So what we've done is we have put in cameras on top of screens and we start tracking on how people consume content and, you know, with that, we call it “content effectiveness” where we have a rating on how people's basically attention towards the content would then indicate effectiveness of the content.

[00:15:09] Out of that exercise, we have helped Westpac and their agencies to make better decisions on what kind of content to put in, what product category they putting in - is it a product base, is it a lifestyle base, is it mixed? What kind of animation style they need to use, size of font, and even day parting on the sort of schedule.

[00:15:29] And after all of that, we've gone through COVID and we've learned so much through it. But the success part is that they've actually managed to increase their conversion rate by almost 87%. So that's really success. Even despite COVID, the kind of number of people going into branches have kind of significantly declined, different personas as well are going to branches, not unlike before where it's consistent and it's predictable.

[00:15:57] So it was, it was really, really amazing to see that project still continue to thrive despite COVID, and despite this sort of a lot of change in terms of how banks are interacting with their customers.

[00:16:11] Bill Pearson: Yeah, 87% is an impressive number. And you can just see the power of getting that feedback on the content like you were describing.

[00:16:19] That's amazing. So where can folks go to get more information about meldCX and Viana?

[00:16:25] Thor Turrecha: To learn more, they could go to our website: meldcx.com/viana. Also there is a QR code they could scan and basically that would redirect them to the right page.

[00:16:38] Bill Pearson: Awesome. Thor, it's been a pleasure talking with you today.

[00:16:41] Thank you for your time. 

[00:16:43] Thor Turrecha: Thank you, Bill.