Advancing Vision Intelligence Filters with Plainsight’s Kit Merker

This is Textron tv. Hey everyone, welcome back here to techron tv. You know, we've got a, an old face, old friend, new company to introduce you to today.

Um, a lot of you know Kit Merker. I know Kit Meer. I don't know, I probably know Kit Mecker 7, 8, 9 years maybe more.

I first met Kit. He had recently come from Google over to Jfr, and he really helped J Rog get up and running in that, you know, crazy DevOps DevSecOps kinda world that existed then. Uh, kit left Jfr, I don't know, I'm gonna say four or five years ago, and helped a bunch of friends start another company that we've covered a whole bunch.

And that's Noble. Nine people kind of brought SLOs to the four, um, in, in the, uh, SRE space. 0.

Um, kit Kit's next kicks Kit's next gig kit. Welcome to Text Drug tv. How are you, man?

Hey, Alan, great to see you. Thanks for the, uh, the history lesson on my career. It's awesome.

Yeah, Well, you know, I saved you from having to say it, right? But, um, but you know, all kidding aside, what, what's up man? What, what, what's the new, what's the new thing?

Well, I'm, I'm really excited. I, uh, just, uh, announced that I'm CEO at a technology company called Planeside Technologies, and it is computer vision, uh, artificial intelligence, ai, um, focused on solving business problems. Uh, if you're not familiar with computer vision, just on that topic, it's actually a fairly mature technology.

It's funny, it's, I think this is the first time I've worked on a space where Gartner has put it all the way to the right of the, uh, hype cycle. Uh, so computer vision, it, you know, we use it all the time in consumer technology when we're, you know, you have an Instagram filter, you have, you know, a, a green screen or you zoom, you blur your background and zoom or, or hangouts, uh, that's all computer vision. And then you see these really wild applications all the time in sort of the news and the future.

And like, one of my favorite, um, solutions is, uh, a tractor that can weeded with lasers. You know, it actually blasts lasers, like 10,000 a uh, a minute or something. Um, and that uses computer vision.

You know, we see it when we're doing face detection, all these different computer vision, uh, applications. And what we're doing at Plain Sight is we're focusing on the companies that perhaps have a, a slightly underserved, uh, market. So people who, you know, own sort of shops and facilities.

And we're working with, uh, marine biologists to count fish under dams. We're working with people who are preparing, uh, food and, and protein part of the agricultural supply chain to, to measure quality, uh, and inventory. Basically the way that we're doing it plain side is we're, we're taking anything where there's a camera and you wanna get a spreadsheet.

I'm oversimplifying, you know, every business, uh, basically is running on spreadsheets. ERP systems are over glorified spreadsheets. Let's Absolutely.

That's always your biggest competitor. That's right. And so if you imagine you have a business where you have physical locations, you wanna connect that to your kind of digital footprint, right?

You wanna say, oh, I've got these stores and I've got these, uh, customers, and I've got this inventory, and I've got these people trying to rob me, and I've got all these different things going on in the real world. How do I bring that into my spreadsheets? And that sounds actually like a pretty hard problem.

You think, oh, I gotta go out and buy some new cameras. I gotta buy some new technology to install that. Um, well, actually we think there's a better way, and it's a pretty simple concept.

Uh, we stole the, i the name from social media, which is a filter, and we call this a vision intelligence filter for, you know, a for complete name. But a filter is what we call 'em for short. And the idea is you can take any camera in your environment, whether it's, you know, a security camera or, uh, infrared or you know, a drone.

And we take the, that footage, images and, and video footage, and we apply filter, and the filters come from this vast repository of computer vision capabilities that are designed to solve a specific problem. So, you know, maybe you're making a pizza and you want to put the right number of toppings on the pizza. So there's a pizza toppings filter, and it will tell you, you know, how many toppings are on it, and if there's too many or not enough, send you alerts, give you the data about it, and now, you know, that can scale to a much larger world.

And the other cool part about this is that the filters themselves are designed to be updated. And I don't know if you know, but I, I know you do obviously Alan 'cause the DevOps background, but like in the world of DevOps changes the constant and AI is changing faster, I think, than any technology, uh, that we've seen in our lifetimes, right? It's moving at an incredible pace.

And so the question for companies is how do they keep up with that, with that change? And so that's really, I think, one of the things we're trying to embrace. So it's really about computer vision.

Connecting AI to BI is probably the best way to think about it, right? How, how do I take artificial intelligence and business intelligence and, and make that work for a business and then have that concept of continuous updates, continuous visibility, you know, being able to make changes in the software and changes in the system without, uh, that being a colossal amount of work or a surprise, something that's kind of a natural part of, uh, adopting the technology. Cool kit.

So Plain Sight Technologies. Look, this isn't a day one startup that you were in stealth and you just emerged and blah, blah, blah. This is a company that's been around for a couple years at least, right?

Um, give us a little of the history if you can here. Yeah. So, so play site technologies is, is actually a new, uh, a new construct, if you will, of what this, uh, you know, historically had had existed.

And the, the technology base that we're building from, you know, it's eight plus years of r and d that went into building the technology, but it was focused on custom projects and focus on, um, uh, really building tooling for machine learning experts. If I'm being honest, that's really what the, the foundation of the company was. It went through some other iterations previously, there was a, uh, it went through its iot phase.

Um, and, and so there, there were, you know, all of those, those kind of phases in this, in this evolution of the technology, the technology now is kind of the day one. And we're, we're now building that as a software, uh, platform and really as a set of software tools and not targeting machine learning experts, actually, it's going to operators and, and business people who have specific business problems in manufacturing and logistics and other, other places. And now they can license that from us with really, I think, a pretty complete set of services to support it.

Um, and it really is a new thing that we're doing, to be honest, which is kind of cool. We kind of hit the, the big reset button and, um, and at the same time, we're able to bring forward the technology assets and the expertise, um, without sort of the, the, the history of, uh, all the other, you know, evolutions that businesses go through. So, Sure.

So kid, look, we've covered a lot already. We're only halfway through the interview. I, I wanna make sure we don't overwhelm people, but for people saying, okay, I think I understand it.

I think I got it'd to, I'd like to just do some of my own reading on this. I want to do my own exploring. I wanna, what, what is the, you know, obviously there's a website.

What's the website and what, what would you, what would you recommend people do there? Well, one of the cool things we have on our website, which I'd say is, is fairly minimalist, uh, and hopefully isn't too overwhelming, is we have the filter repository. com, either one, um, you can click on the explore filters button and the, the filter listings that are on there give you, you know, some examples of the kinds of things that can be done with filters.

And you'll see that there's actually quite a wide variety. The interesting and, and hard challenge in building this type of technology is that vision is everywhere. It's kind of our most important sense in a lot of ways, right?

What, what you see is what you get. And so being able to target, you know, this anything concept, right? I can look at anything.

I can make data out of it, I can make sense of it. Um, but making that trackable, as you said, like trying to make it, you know, not overwhelming and understandable, the way that we've, we've kind of, uh, helped with this is creating this filter repository. You can go look and see, you know, just like an app store, basically, right?

You can see all these different solutions in there that do everything from, you know, privacy shield, which will blur out faces to wildfire detection, which is an area we're very excited about to help, uh, with wildfires. And, um, hopefully, uh, help, uh, you know, give back to our, our firefighters out there in the field. Um, we have underwater, we have, uh, you know, for marine biology, we have, uh, things that can read labels, things that can count sheep.

I mean, there, there's so many different filters that are available. And so if there's something that matches, you know, a business need, we can help get that to a customer and help them, uh, deploy it. Um, at the same time, you know, if there's a, a use case that isn't covered by the filters, we're happy to, to talk about it and see if we can find a way to build it.

Um, and, and oftentimes, um, we can, we have a pretty strong library of general purpose computer vision capabilities in terms of, you know, counting things, looking for faces, uh, looking for license plates, um, measuring wait time, measuring abandonment, and drive-throughs. I mean, there's so many different things we can do that help businesses, uh, with filters. And it all starts from, you know, figuring out which, which filter makes sense to as a kind of an entry point.

And then once you have that deployed, you say, okay, I've got some cameras. I get a filter running, I configure it, uh, and then I get the data I need, that data starts flowing to the right dashboards or reporting systems or automation. Um, then we can start to expand that and look at how do we scale that up to a larger set of cameras or potentially, you know, each department might want a different set of filters.

This is one problem I've, I've actually seen quite a bit, okay, I've got a, a, a retail space, I have cameras in there, and different managers wanna watch different things. Some managers want to pay, pay attention to the employees. You know, do I have enough, uh, pieces of flare, uh, so to speak on my uniform, right?

Am I, am I in compliance with the, the uniform guidelines? Am I, you know, vaping in the hallway when I shouldn't be? Then you have different people who are concerned with maybe a VIP program.

We wanna identify customers that are likely to spend money, let's go find them. Then you might have somebody else concerned with security and loss prevention. Well, if you, if you layer on all these different interests onto the camera, pretty soon it's just clutter, right?

It's like, okay, now I've got computer vision, but it's not really telling me anything. 'cause it's just looking at everything. The, the plain side approach is you actually create different filter pipelines for each of those different constituencies.

So you can customize and update and also get alerts that are targeted to your function. So this is a, this is a killer capability for organizations because, you know, you can imagine the other alternative, you buy eight different point solutions. Now you got an array of eight cameras in each place.

'cause you know, there's a camera for the security guy and there's a camera for the, uh, the, uh, customer care person. Uh, so that's a pretty cool way to solve that as well. But, um, yeah, I think it really does come down to trying to solve a business problem is the first step.

And if you, if you have a business problem, you say, okay, I'm paying somebody to stand around and watch this and write it down or take action. That is something we can explore. And the advancements in computer vision now is kind of a solved problem, actually, the computer vision problem itself of like, okay, can I point a camera something and extract meaning out of it?

That's pretty solved. The harder part is how do I get started without machine learning expertise? What hardware do I need?

Um, and we are actually sending hardware starter kits to all of our customers. So we send them, uh, uh, uh, Jetson Odin device from Nvidia, which is this GPU powered device. They get to use it for the evaluation.

If they buy from us, they get to keep it. If they don't, they can send it back. Um, and that's helping us accelerate, uh, you know, getting started.

A lot of cool things that can happen there. Uh, but getting past all of those little hurdles so that you can prove the concept and go from idea to industrial scale, you know, in kind of this very smooth way, which is what we've seen in the cloud with DevOps over the last however many years have been doing DevOps. Uh, now we're trying to apply that to this, you know, bringing AI to the edge of a business where they have a physical space, connect that back to their data systems and make it easy for them to manage that over time as AI going.

And by the way, I think the other key point, um, is that a lot of AI companies are stealing data and harvesting data from businesses to create products, right? They think that their IP is gonna come from all the, the models that they're able to build. I think that in the business context, we can talk about consumer privacy.

I think that's a different issue. But in the business context are, uh, the, the, you know, the, the data derived from our operations and our cameras is pretty important. Competitive, you know, intellectual property.

And if you, you know, you would not be happy if your business, you know, you turn around one day and like the AI company you bought from was selling that AI model to your competitor and they trained it all on your data and your brain, you wouldn't be happy with that. And so one of the things we've taken a stand on, I'm not, Yeah, right, exactly. Yeah, you would not, no one, no one would be happy with that.

So, but I see this pattern a lot in the AI market. I see companies that are, you know, harvesting data from their users sometimes many times without their permission and definitely without their compensation. And so one of the things we've done with Plain Sight that I, I'm very proud of and I'm, I'm quite, uh, excited about, is we will not harvest customer data for models that we sell without full permission and or compensation.

It is just against our policy, uh, against our way of working. And because we have this filter concept, we can decouple the filter from the model, and instead of selling you a model, I'm selling you a filter, which is a fully blown application. You can actually bring your own model, we'll help you train it on your data to fine tune it, make it more accurate, but you also get to keep it.

And that privacy issue is so important, um, that I really think it's, it's a key differentiator for us because I don't think other people are doing this. So the ability to have that very powerful ai, very customized and tuned to your environment, build up that ai, uh, you know, intellectual property over time from your models, and then also having the ease of something that feels like deploying social media or a Zoom where it's like, it's just a filter. The filter does what you need.

I think it's a powerful combination. I'm very excited to get this, uh, started and to help out more customers with this, uh, technology. Absolutely.

ai was one, one, uh, URL, what was the other one? com. Either one works.

So if you, you know, if the AI isn't working for you, plain sight is spelled like, you know, the AI is in plain sight, so it's like the way to remember it. There you go. Yeah, that makes sense.

Very cool. Like planes. Alright.

Hey Kit, first of all, man, congratulations. You know, it, I think they made a good choice bringing a good CEO in here. We are gonna be following it.

And you, uh, what you're doing at Plain Sight, I'm sure it's gonna be a great, uh, story. ai. We will have a little bit, it sounds like this's the DevOps application, and I, I don't know if we mentioned this, is, this kind of runs in a Docker kind of environment, so it can run anywhere you run in Kubernetes and Cloud Native.

And Look, it's the, it's the 10 year anniversary of Kubernetes. You gotta run in Kubernetes. It's taken over the world.

But, uh, yeah, so, so on the technology side, yeah, filter, it's a Docker container and that means, like you said, it can run anywhere. It runs in Kubernetes, it runs in K three s, which is Kubernetes at the edge. Um, this lets you have ultimate flexibility.

This is not a SaaS offering, this is a, you know, it's a real software, uh, business and we make it easy to deploy and update because Docker is great for that stuff. So, um, it makes it teams happy, it makes security teams happy, uh, and then the end users are happy 'cause it just works and they don't have to think about, you know, all the complicated installations. So it, it really is, I mean, it's amazing for me personally to come full circle 10 years later from working on Kubernetes to now be using it as like a key technology for our, uh, deployments.

It's kind of a, a cool thing. Absolutely, man. All right.

Best of luck, kit Meer, CEO, plain Sight Technologies here on Techstrong tv. We're gonna take a break. We'll be right back.