This is a recording of meldCX's session "Drive premier customer experiences through AI & intelligent edge technologies" at Meraki Fast-Forward: Make every space a sustainable smart space event in Singapore, 6th July 2022. Brought to you by Cisco Meraki & meldCX.

‍

The future is here and sustainability has become a critical strategic objective of businesses large and small. With this focus, IT leaders no longer ask, “Why sustainability?” Instead, the focus has shifted to facing the challenge of “How can I achieve sustainability and continue to accelerate productivity and maximize ROI for the business?”

Learn how you can redesign your customer experiences with AI and intelligent technologies, powering up amazing use cases you never thought possible.

Featuring:

• Joy Chua - EVP of Strategy & Development at meldCX
• James McKee - Meraki IoT & Edge Intelligence PSS (APJC) at Cisco Meraki

Transcript:

[00:00:00] Joy Chua: Thank you everyone and huge shoutout as well to everyone who's joining us virtually as well, can't forget you guys who are joining virtually. I know some of my team are on as well, so wanted to make sure I say hi. My name's Joy, part of the meldCX team. I look after strategy and development globally for the business.

[00:00:18] You know, a little bit, but meldCX as well. We've been a strong ecosystem partner, I hope with the Meraki team. And what we do essentially is to deliver AI use cases that, you know, people thought were never possible, but we make possible on Meraki. So meldCX actually stands for melding the customer experience and to a Charles' point as well.

[00:00:37] I think I was set in on his conversation. He said, you know, making sure that whatever you deliver is outcome based and outcome focused. And that's what we do with our customers. We work with a large range of partners as well. You can see on screen up there, front and center, Cisco, Meraki. We also work with the likes of, you know, Google, Intel, and Microsoft as well.

[00:00:57] And we've got, you know, our channel partners to o so I think the topic of smart spaces we've talked a lot about as well. We're really keen on smart spaces, I think. One thing for us is to make a space. Is to really make sense of the space, and that's what we do and that's what we're passionate about as well.

[00:01:15] It's all about, you know, making sense of the space, providing a platform for people to understand the analytics of that space and surface them up into dashboards so that it's really consumable and all leads to a specific outcome for customers after. So essentially what we do is to be able to give customers the opportunity to track physical spaces like they would a website.

[00:01:37] So that's what we do.

[00:01:39] James McKee: That's, I like the alignment to website cause it really makes sense. Yeah. So, so we've been working really closely together on a range of different you know, problems for our customers across across the region and, and globally. I guess, you know, the way in which you actually you know, present a lot of that data is, is using your platform called Viana.

[00:01:54] I was wondering, I was wondering, would, would you be able to, you know, share Viana with, you know, with, with, with the audience today?

[00:02:00] Joy Chua: Definitely. So Viana is the product that we've been working really closely on. It's really an end to end platform that's up and screen as well. We've built an end to end platform that's all about helping organizations maximize the understanding space, visualize data through actionable insights and help hit their objectives and outcomes as to why they're having, you know, and using that AI use case. So we'll talk a little bit about some customers. It's actually a joint customer that we've worked on based in Australia. So we'll be sharing a little bit on that.

[00:02:27] James McKee: Definitely. So, so on the topic of Viana you know, I guess if we think about some of the customers that we've been working, You know, the problems are wide and varied.

[00:02:35] And you mentioned defining the problems really important, and then actually, essentially applying the toolkit to you know, to solving that problem. Especially with say things like cameras and, and, you know, techniques like computer vision. Could, can you maybe explain to us how you, you go about actually applying that?

[00:02:49] Joy Chua: Definitely. That's actually a question that we got asked quite. Just outside on the booth as well, people asking us, So how do you actually do it? Where does Meraki actually play into it? What do you guys use of it? So we've come up with a little diagram as well. Hope it shows up on screen as well. And this is what our process is.

[00:03:06] And essentially we've coined it and we've given it a term, and, you know, mapped it according, I guess, to a sensory process as well as to what it is when a human brain recognizes and makes sense of stuff. So we've called this the Viana cortex, and that's essentially what we do. So in this scenarios, you can see on screen we are looking and using the MV cameras, the Meraki cameras, essentially as your eyes.

[00:03:31] So we take the RTSP feeds from the cameras. We take that, we process that through a range of services, which could be, you know, either on the edge or the cloud. We've got two options available depending on the use case we need to track, we process that through services on Viana and the processing.

[00:03:47] And, you know, what happens on that AI perspective comes into that Viana cortex where, you know, that's where all the algorithms are placed. So that's where the processing happens. And then the visualization is the output of the data. So that customers can actually take that, make sure that, you know, it fits into their outcomes and make them necessary pivots or human interventions needed to deliver that use case. So that's a total overview of what we do.

[00:04:11] So we've got the perception stage of which, you know, we work really closely with the Meraki teams, the cognition stage, and then finally the decision stage as well.

[00:04:20] James McKee: That's, that's awesome to get that overview. And, and it's always exciting to have these conversations with the customers and actually see how they actually apply these tools.

[00:04:28] We've, we've talked a lot about data today and I think, you know Charles really high highlighted the importance of, of data in the whole process. And I guess I tried to kind of un underpin that conversation. I was wondering if you could maybe drill into a bit more detail around the platform and, you know, if you could highlight some of the key messages that you know, or takeaways that, you know, everybody here attending should be mindful of.

[00:04:47] Joy Chua: Definitely. So we've put that to five key use cases as well, so it's up on screen. We've essentially, you know, really drilled into what it is to make sense of the data.

[00:04:58] And I think, you know, we've spoken about this, making sure that data's not in isolation. You know, Charles talked about making sure that data makes sense and enabled humans and so. You know, we have done that as well. And so one core thing that we do is point 2, but I really wanted to focus on that is making sure that insights that we get are meaningful and addictive at the same time because we wanna make sure that data is consumed and continues to be consumed.

[00:05:23] And I think that's where, I think in Charles' conversation he talked about having a digital twin, and that's really important to us as well. I think, you know, one thing about having meaningful and addictive insights is to make sure that it's surfaced up in a way that's consumable for everyone.

[00:05:39] So for example, you know, if I was a marketer and I wanted to know, what, you know, customers were interacting with in my space that, you know, I was able to access the data without having to go through that whole circle of reaching out to business analysts. You know, I had that at my fingertips essentially because it helps me do my job better without having to wait for things to come through.

[00:06:02] And so that's what we've done. We've really taken the effort to make an easy, comprehensive dashboard with the right UI so that customers could do that so everybody can do their job better. So that's a key thing that we really focused on. Another thing that we do is what we call anonymous re-identification.

[00:06:19] So that's I guess one of the key value prop we always talk about as well. You know, to make sure that as we are, you know, tracking anonymous personas as they go about the store, for example, I'm just gonna use retail as a general example cause it's quite easy to understand retail.

[00:06:33] We are performing what we call anonymous re-identification. So in our process of going through AI and ML, we've learned really early on that, you know, to basically group and get good insights from a certain space, you actually don't need to do things like facial recognition or, or anything else.

[00:06:50] You can actually have a mixture of different data points to add breath to what you want to basically anonymously monitor. So things like what you are wearing, for example, James. So if James say were to walk into a retail store, we would pick up, say , male aged within 20 to 30,

[00:07:11] James McKee: I'll take that

[00:07:12] Joy Chua: 20 to 30 wearing a green. You know, he's interacted, say with a couch. Then he's interacted, say, you know, went to the self service furniture section and then ended up at the cafe. Cause he has to wait for his partner, maybe. And that's, you know, his journey. And so we group all of that into personas and that key value point as well is about anonymously reidentifying someone.

[00:07:32] So James in our system will not be known as James. He'll actually be given an anonymous token, like a 003. So it'll be 003 with all these data points walking around the store, understanding how he's interacting with the space, surfacing that up into data, and then giving that to marketers or sales reps to understand and cater to how we should scope out a space, especially if, you know there's little retail space and we need to maximize that.

[00:07:58] Some other things as well include things like minimal bandwidth usage and I'll put ease of deployment together with it as well, because that's really important. So once you know all your Meraki cameras are set up, we have ready to go modules that are available to be loaded straight onto the cameras and deployed that way. So you don't actually have a very lengthy process to do that. We've made it really simple so that anyone can really access the data as effectively and efficiently as possible.

[00:08:25] And then the last one I know is something else that we're really passionate on.

[00:08:28] James McKee: Yes.

[00:08:29] Joy Chua: AI on cameras, which is a latest feature in the Meraki ecosystem called Custom CV . And we're really honored as well to be part of the first few beta deployments for it. So essentially what that is, is, you know, we're able to anonymously track how people interact with spaces and have that straight onto the cameras.

[00:08:47] So you know, we don't really need to use any edge devices. It goes straight on there so that it expands what we can do on the camera. So we're really excited for that.

[00:08:58] James McKee: That's, yeah, I think that concept in itself, that whole edge AI processing piece, I guess it really opens up the opportunity for customers to deploy these technologies that often had traditional barriers to entry.

[00:09:09] Which I guess is super exciting. That's probably one of the biggest inhibitors is actually getting the technology in place to drive that.

[00:09:15] Joy Chua: Exactly. So we've been working, I can't really say the customer name right now, but we've been working with that deployment has actually been rolled out in the US but we're using that for specific large multinational coffee company and we're using that for their drive thru.

[00:09:29] So we're using that to understand, make a model of vehicle and how we can help with drive thru processes. So making sure that as people drive through to pick up their coffees, you know, we're keeping wait time low. We're doing stuff like predictive analytics. If someone drives up, say, you know, the office junior comes up to pick 20 coffees for a large board meeting that, you know, we see that same car comes through, we can start, you know, making oat milk ahead of time or making, you know, in America they've got half and half, so like making milk for half and half because that order remains consistent.

[00:10:04] So that's some things that we're working on with custom CV, which is super cool.

[00:10:09] James McKee: So you just reminded me if we think about CV and custom cv, I know you also do synthetic training in custom model development. And that reminds me of, I guess one of our large postal customer in Australia.

[00:10:19] So maybe you could share a little bit about that customer and how they're applying your technology to drive success.

[00:10:24] Joy Chua: Definitely. So I think that's a really, that's a customer that's really close our hearts, I think because we've worked on that together in Australia. So this is the customer that we've worked on in Australia.

[00:10:35] They came to us with very clear business problems and key outcomes that they want to achieve. So for this particular postal network, what they came up to us was, you know, hey meldCX, hey Meraki, we've got a couple of things that we want to achieve. We want to digitize our post office network or, you know, postal network.

[00:10:56] And there are additional things that we want to understand. So the first thing that they wanted to understand was how are customers interacting with the idea and concept of parcels? Has that changed with the pandemic? So in Australia, post offices were consider an essential service. So they were open for the whole duration of, you know, the lockdowns and, you know, everything that happened in kind of Australia and Melbourne and I'm sure in Singapore in various parts of APJ as well.

[00:11:23] So they wanted to understand how passive behavior had changed. Do we want to walk up to a kiosk? Will we rather have human interaction? Or you know, can we just drop something off a parcel schute? Like what do people want to do moving forward with parcels? So that was one.

[00:11:39] The second business problem they wanted to understand was, you know, we've got x amount of floor space and floor space is becoming more expensive year on year. How can I understand what elements of a postal network or the retail facing element of the postal network needs to be? So do people want more letters to pass those? Do I need to have a bigger area for passport taking for, you know, vehicle registrations? What can I do? So they wanted to get more information on that.

[00:12:09] And then the last one was understanding foot traffic, people count to renegotiate leases as well as do things like understand opening hours. Do I need to open on a Saturday or Sunday? Or should I start later and end earlier? What are my best times?

[00:12:25] So that's what we worked with them on. And I think this customer is really dear to our hearts because they use a wide variety of AI use cases all on that same platform. So if you see on screen here they use all those use cases that are up there. So they do things with us on that same platform for zone engagement, audience measurement and content effectiveness.

[00:12:47] So zone engagement is you know, essentially allowing customers to understand dwell times, key personas that are interacting in zones, you know, who comes into a postal network, retail shop the most. How long do they dwell there? Do I need to expand floor space because of that? So that's one.

[00:13:04] And then, we also do things like content effectiveness, which is really closely linked to digital signage and making sense of what's on the digital signs. So does it cater to them, do more people look at forex ads instead of passport ads , all of that. So we're really keen on this customer and they use all of these on that same platform for those outcomes that I mentioned before.

[00:13:25] James McKee: That's awesome. So could we actually see what this platform looks like?

[00:13:29] Joy Chua: Exactly. So I'm gonna walk over to my laptop to give you guys a bit of an overview of what's available. So hopefully we can swap the screen over to the other input. Awesome. So that's come up. So essentially, I hope that's big enough. I might make it bigger for everyone.

[00:13:51] So this is what we meant by making sure, and I know there was a lot of, if you guys were here in the AM session, there was a lot of talk about making sure that user interfaces are really easy to use. Bear in mind, I have a marketing background. I'm actually not technical, but I can use this platform pretty easily.

[00:14:06] So all the things that I mentioned before, including things like people counting and zone engagements, I'm gonna show you how easy it is to deploy. So, for example, if I go here, this is actually one of our regional offices as well. So the team knows that this is happening. I hope that loads up.

[00:14:27] So essentially I wanted to show you guys how easy it is to basically draw a tripline. For some reason it's not working. I'll come back to that in a tick. Hopefully it loads up. I also wanted to show you guys, this is an extraction of the dashboard that's available. So if you can see on screen here, this is a, you know, ready to go dashboard of all the elements that can be anonymously identified and tracked for a space.

[00:14:50] So this is assuming that it's a retail space. You get things like gender, demographic , ins and outs, total visitors, who's spending time in zones, over capacity incidents, and best times.

[00:15:02] One of the key things as well which I mentioned about the post office network. This one, they love this the most.

[00:15:09] Understanding the top three zones and making sure that they were able to maintain those zones, expand those zones, and make sure that they had the best service for those. I'm going to go back here, awesome, it's loaded up. So if you can see here this is for a use case for people counting. It's so simple. All you have to do is once the Meraki camera's up, this is showing the entry.

[00:15:29] All you have to do is just draw a line. This is a tripline as you can see you can label this, you can change the name of it, you can put a doorway to it, and it's also directional, so you can swap it. So this is going into the door, this is coming out of the door.

[00:15:43] The last one that I wanted to show as well, I'm gonna discard change, is how easy it is to do a zone.

[00:15:49] So, for example this is, sorry, I might give it a while to load. It's up here. So this is an area, this is our actual test area. If you wanted to start getting some information on zones, all you have to do is just draw a box. You can manipulate it like you would Photoshop, save it, and it starts tracking straightaway.

[00:16:11] So as you can see on the screen as well all our faces are blurred on stream. So, you know, we're not doing any identification. And we've got an extensive list of privacy. And if, you know, I don't think I've got time to cover it off right now, but if you guys wanna know more about what we're doing for privacy, GDPR, please have a chat to us.

[00:16:30] We deliver the solution in North America as well as in Europe. So we've got a whole set of things ready to go as to how we are dealing, storing and transacting in data. Awesome.

[00:16:44] James McKee: Thank you Joy.

[00:16:46] Fantastic overview. Following some of the applications that we've seen in our region where we apply this technology within our customer environments to drive success.

[00:16:53] We're just about at the end of our time slot for this. So what we might do is hand back out over to the crew and get ready for the next session now.

[00:17:01] Joy Chua: Awesome. Thanks guys.

[00:17:02] James McKee: Appreciate it. Can I have a round applause for Joy? Thank you for sharing.