Video: How To Get New Insights From Surveillance Video With Kibsi

hey Channel insiders welcome back to channel Insider partner POV your source for untapped opportunities and unfiltered opinions in the it Channel all to help you make Better Business decisions I'm your host Katie boso before we get started have you subscribed to our YouTube channel yet head on over and check out our page @ Channel Insider news and Trends my guest today is Tolga tarhan CEO of kibsey a startup on a mission to make computer vision lightweight cost effective and accessible to everyone customers can even use their existing camera systems to collect data and insights like never before Doo welcome it's great to see you thanks for having me excited to be here thank you so much I'm excited to have you here because just browsing kib's website uh which I've obviously done more than that but anybody who just goes to your website immediately sees that you do something I think maybe 10 years ago we thought was a little bit more James bondes than it is right now with computer vision uh so why don't we start out by talking about what kipy is when you describe it to people how do you describe what kibsey is the solution that you offer and the kind of Partners or I should say clients that work with you yeah so kipy turns what we see in a camera feed we turn that image into Data in real time and so it's a little bit like Minority Report you know where you're just analyzing what's happening in the real world but think of it as business intelligence powered by your cameras today you've got an Erp system or or a warehouse management system and these systems are actually if you think about it secondhand sources of information right like they get their data because somebody enters it or because maybe there's a barcode scanner along a conveyor belt or something of that nature but nothing is looking at the actual human activity and measuring that human activity and so I think one of the really neat opportunities is to say hey let's use these cameras that are already deployed as an asset throughout your organization and let's analyze business processes and let's say hey here is how this process is working we've built a platform that lets you express those kinds of business problems and then delivers data from your cameras so I want to back up for a moment just in case we have some folks listening or watching who may be new in this space can you define computer vision for us it's a type of artificial intelligence right so there's lots of kinds of AI right now the world is dominated by kind of generative AI chat Bots those kinds of that's like all the talk today right but there are other fields of artificial intelligence computer vision being one of them and there's lots of techniques in computer vision so it's a whole category unto itself and it's also evolving very quickly with all the recent innovations that are happening in AI at the heart of it though is we take again images from cameras that are really not structured we take that kind of image and we turn it into hey there was a person here doing this thing there was a box here with the stuff in it it's kind of taking apart what we see in the scene and making it into something we could then store analyze as if it was kind of Digital Data to begin with yeah and just to give the audience an idea King research estimates that AI computer vision that market will experience a lot of growth between now and 2030 growing at a cagr of more than 37% so we're talking a market that is very exciting you talk about AI toga that everywhere right now that feels like an understatement honestly because it is just so intrinsic and and ubiquitous across the technology Channel and the IT industry in general but as far as bringing this into a human perspective some of your clients we're looking at a lot of big spaces like airports so transportation in general we're looking at venues so stadiums and we're also looking at one of my favorite Parts zoos which I thought was really exciting zoos and Aquariums but can you give me a real world success story in which kibsey was able to bring in this awesome technology and be able to change the way or improve the way that client does business I'll start with Zeus because you know we spent a lot of time in our business looking at camera feeds of industrial environments like a warehouse or a factory those are interesting places but they're not nearly as fun as looking at camera feeds of cute animals at a zoo so our team really likes when we work get to work with zoos just because the the subject matter is more interesting more visually appealing so in zoos what we're doing is helping uh with Animal Welfare by monitoring the behavior of animals so in let's see have a large enclosure with elephants in it you know where are those elephants spending their time in the exhibit are they showing behaviors of maybe being too anxious or anticipatory these things help uh catalog the behavior of animals to ensure that the animals are well taken care of that they're well that the enrichment is working that that they're not bored or or or anxious we've been really lucky and fortunate to have a handful of zoos that were innovators in the space and essentially what we provide is an additional input to their Animal Welfare team that they couldn't really do before right because you can't have a person sit there 247 and watch the whole exhibit I would do it I'm just saying that if they needed to volunteer I might do it a sleeping bag sleeping bag in a little campsite right outside the exhibit all the El to your point they do use volunteers to do uh sort of measurements and little snapshots for the next week let's let's spend eight hours a day um you know recording this data but that's a snapshot and it might be a while before you get the next snapshot and so this is a way to do it more continuously but most of our customers are industrial customers so they are um factories warehouses they are creating product or or storing product or shipping product you know I'll give you an example of a large you know construction material manufacturer you know you can might might imagine conveyor belts producing products that are going to be used by construction workers there are different things that could go wrong they it could be bundled incorrectly meaning maybe it be a pack of four or five but that didn't get packed correctly all these things are going to cause a machine later on down stream to to jam and they're going to cause a stoppage in that line even if it's a very momentary minute or two to fix it if that happens every day and it happens across say 20 facilities there can be a big impact right so what you can do is you've got really three options you can put a lot of humans on the problem um which is Impractical you can uh use really specialized kind of Industrial Automation and rework your conveyor belt and rework your line to deal with this problem right using very very specialized Hardware um or you can aim commodity cameras that are probably already in your ceiling at the conveyor belt and then use technology like kipy to get almost as good as that second option but without having to stop and redesign your production line and so what kipy does is lets you address these problems that are paper cuts that are causing an inefficiency they're causing an outage they're causing a production problem and it lets you do it without investing millions of millions of dollars in reworking an entire production line just aim a camera Define the business process and start getting data out so you mentioned also that you can use the existing camera system when we say that I mean what are the stipulations there I mean how how easy is this to kind of click and deploy in a system when we talk about using Kipsy most of our customers do already have cameras that were installed for security purposes right your average sort of Enterprise environment has relatively decent camera coverage and all you're doing with that footage is recording it um storing it and then you look at it if somebody gets hurt otherwise you delete it you know a year from now that's the typical lifetime of this footage well we take those same cameras we leave them alone from a security point of view and we process it in the way I described earlier and so really it becomes very simple to turn this on because if the camera is already in place it's all just software right there's no there's really no infrastructure to go change and then what we've also found is in the event that a camera's not there our customers typically already have a vendor they're using for cameras where they do have them and so it's easy to add one to that system so we're not coming in with a hey here's a major new installation project it's more um here's some software that basically gives your existing infrastructure new capabilities I think it's a really compelling way for customers to start adopting AI one of the things I like to talk about a lot is Enterprises are all looking for hey how will AI affect us and you know will we get disrupted by AI or can we be a disruptor with AI and I think Executives you know in manufacturing in it in supply chain are thinking about how do what how's what's it mean to me how do I become the person who brings AI into the company in a way that's impactful and that's where kipy is so powerful because with just software with just essentially point-and-click kind of work you're starting to Leverage The Power of AI and you're not having to do years of capital expenditures or years of R&D and I think that's really where the power comes from and on that AI note was this something that you saw as as possible as it is now three five years ago or has AI really been the engine for the way that kibsey can exist as we know right now uh the way that AI has been on just such a upwards trajectory in the last 18 months it's definitely new tech right like I think the foundations of this go back to the sort of mid-2010s like that that's sort of where the foundations of modern computer vision start but it had to get cheap enough and fast enough and reliable enough and then you needed some someone like us to come around and realize hey we can't keep doing custom Professional Services projects each time a customer has a need in the space there's a need for a platform there's a need to say how do we build these things without millions of dollars of R&D every time repeating ourselves and so our real innovation in the space isn't the best new computer vision models in fact we don't see that as our role we see our role as taking all the great models that exist in the world or all the great techniques that exist in the world put them into a platform and let customers adapt them and apply them to a real use case but we didn't see was there was no no real way to like apply it to a real world business problem in a way that connected with you know a VP of supply chain a VP of operations a COO not a computer scientist right and so that that was the Gap we filled I'd love to know do you have any benchmarks that you might be able to call out be that where a customer was either able to save x uh amount of money or perhaps increased productivity by X because of kibsey we're relatively young still so it's hard to get like real tangible metrics what I can say is as a customer of ours that's using kipy to manage the loading of uh of trucks and in that scenario you might imagine there's a lot that happens kind of from you know where the conveyor belt delivers the package to where the package gets onto the truck and in that scenario you might imagine that it's easy for inventory to get misplaced like it doesn't get on the truck either intentionally or accidentally those are the kinds of like really tangible losses that you could could have cuz now where'd that box go depending on what that box is it could be very valuable and so you know I don't have hard numbers but what I can say is we've got customers that lose multiple boxes a day in this scenario and so it's like well how do we stop doing that and we've heard that from from lots of customers it's not just one um in fact it tends to be relatively prevalent anytime you're loading a truck not by the pallet but by the box boxes are much easier to misplace in this process and so you could imagine really easily these boxes being worth thousands of dollars of product in each box yeah uh and you could imagine if you lost even four or five of them a day which I think is low from what we're hearing it starts to become a real meaningful number over the of months and years you mentioned that kibsey is Young and I should make sure I I call that out kibsey really went a market uh last summer I believe in June 2023 is that correct yeah all in all about a two and a halfy old company but it took a really long time to kind of crack how to do what we're doing we have a really ambitious platform and so we spent a lot of time building it and then we spent a lot of time exploring the market so we didn't actually go right to Industrial use cases as the as the first thing and then we realized where the most value could be delivered the most quickly was in these manufacturing and warehousing environments and so that kind of focus and go to market strategy started for us in yeah in June July uh of last year and we it didn't really start to return results until like September October of last year you have a background in Cloud you were the CTO of Rackspace prior to being the CEO of kibsey can you talk to me about where you said there needs to be a product like this out there what was that aha moment for you and what was the process like after that I have a software engineer background and then and then became a cloud infrastructure expert um AI is is relatively new to to me um definitely isn't my background but I think that's what special here by the way is people that think about AI every day have been thinking about it from a very like how do I make the next best deep Tech solution here and we came in and said how do I make this deep Tech accessible because we're kind of software people tackling AI problem and so I think that difference in Viewpoint actually was really powerful and it's the reason our platform is designed the way it is as opposed to being another incarnation of sort of this deep Tech the idea came out of a bunch of work we did in iot so back at Rackspace and the company that I was at before Rackspace anuka which was acquired by rspace we did a ton of work in the Amazon space with iot so we helped customers adopt AWS iot services in that world we kept running into use cases that would be really interesting with cameras but also kept running into a problem where using cameras to deliver them was too expensive not because the cameras are expensive because there was no platform required a ton you know it was like Hey okay we can do that for a million dollars we once had a really large oil and gas company come and say hey we want to do safety compliance we want to know if people are wearing the proper vests and hard hats and we looked at our team and we said hey what would it take to build that out and you're like well we got to build and train the models we got to build all these uis to consume the output we got to create reports and we came back with well it's a million dollars and we we kind of got laughed out of the room saying we're not going to spend spend a million dollars on that project right well at kipy that that project might be you know 5K a month 4K a month right subscription kind of price and so it's a very different way to solve those problems and I think after seeing that pattern a few times and realizing the amount of kind of repeat yourself you have to do to build these Vision Pro products we said all right there's got to be a way to say we're going to take all that heavy lifting and replace it with a platform and instead of it being 100% custom we're going to make it 90% platform and 10% custom how long from when the ink is dried where the customer says yes we want this to it's fired up deployed and installed what does that timeline look like in kipy we have a library of things we can detect like a box a person a forklift a car a truck right if it's in that Library already it's literally hours as soon as we have access to your camera stream we can start to produce the results for your use case if we don't have a model for your use case then it's there's a few weeks of of building and fine-tuning a model and that happens when the object in question might be custom object like if you manufacture pens and you need us to detect your specific brand and make of pen then we're not going to have that in the library right that's a pretty unique object that's a long way of saying it's somewhere from a couple of hours to a couple of weeks let's talk about who is able to support you in bringing kibsey to the masses uh what vendors or or hyperscalers are kind of the engine behind kibsey yeah we are an all-in AWS shop uh my background from the anuka and RX space days was AWS so I've spent probably from 2013 essentially focused on the ABS platform right initially as consultants um and managed service providers and then now as a SAS company building on the platform so our whole team loves AWS as a platform to build on we also love AWS as a as a channel partner their partner program I think is pretty unique and uniquely friendly to startups and so as a company with no revenue and a few people in a PowerPoint you can still get support financial support marketing support and that's been really really cool and so so we've got a great relationship there they not only uh Power our technology but they help us with go to market and they helped us as a startup you know in early days when you're just building and you're having to spend a ton of cloud consumption just to build your product Amazon can be very helpful in those situations as far as with the rest of 2024 what are some of your plans there what are some of your goals to accomplish our main focus now is some scale so you know take the Dozen customers they're working with today and find more that look like them and grow the the customers that are engaging with and also grow the use cases that our customers are using us for we we're a seed company so the name of the game is prove to the market that your product is relevant prove to customers that it can deliver Roi and then go and raise your next round and do it again right so from a tech point of view we feel really good about where we are and from a go to market point of view we have incredible logos with Incredible sort of name brand customers and so now it's a matter of scaling that out and going and getting more and more of it i' also love to know how can you overlap with Channel Partners like msps and vars and perhaps you don't and that's just as much an answer but is there any way that you can interact with them plan to interact with them or that they can reach out to you to be able to drive business outcomes for their customers how does that overlap yeah we think there's actually really interesting Channel opportunities that we haven't really dug into yet I I can list a few so for example any SI right we're really natural fit for an SI because what we do is produce a new source of data so if an SI is on some kind of broader project and cameras could enable a new source of data for that project that's super interesting some of our investors are SI and so that's we're experimenting with that already also the camera ecosystem who makes cameras who installs cameras right so there are vendors who are typically sort of security vendors that install cameras configure cameras configure the the recording devices well those companies could now offer a new product in their portfolio and say Hey what if I make your camera Smarter with AI and so that's another really interesting channel for us so as we grow we'll have to dedicate some energy and some some headcount to working through those channels in the meantime we're being opportunistic as we run into them this is something it's so interesting hearing your story and doing research on you and seeing your background and even you saying things so humbly like ah I wasn't an AI guy you know I wasn't an expert on it but here you are having launched your your business that is so differentiated in this space what lesson can you impart on other uh disruptors who are thinking about about going out there starting their own business with an idea a hope and a prayer I think my my best advice to everybody would be to really maybe hone in on that Target Market earlier I'd focus on the market first and the tech second I think this is probably advice that's been written about in lots of books and lots of entrepreneurial kind of texts but it's really hard as a technology back on person to really appreciate its importance and so I guess you know coming from technologist a technologist I would tell you as a new entrepreneur hey make sure you understand um where that Target Market's going to be who the buyer going to be and what their needs are and and hopefully what that lets you do is skinny down kind of V one of your product to just that and I think that that gives you a better Head Start and sort of a better running start on the more positive side the thing that we did really well is I would surround yourself with people that are better than you when I said I'm not an AI guy so what did we do we went and found people that are much better at that than we are to be on our team to be part of this journey and so I think a good leader uh should look around and say hey all these people are better at this than I am uh I think that's the thing did right and I would recommend you do as well so if we'd like to reach out to you if I'm a customer if I'm a business owner if I see this uh need in my business how can we reach out to kibsey and get started with you yeah so ky.com is the website there's a contact us form there we we read them all they they show up in slack I see them um feel free to mention the podcast and I'll connect with you directly um we're a small team we'd love to chat if anyone wants to talk to me I'm very accessible we are I would say part of a of a revolution here A disruption here in how do we bring applied a AI to this industrial environment that I think has been un underserved in general by the kind of Silicon Valley Tech Community and so if you are in a position where you want to start bringing that Tech into your industrial environment and be a disruptor we're looking for those kinds of customers that want to partner with us and shape the future of this product and shape the future of this approach amazing thank you so much Tolga it's been so interesting to talk to you and I wish you the most success this year thank you thanks for having me thanks so much to toga for joining me today and thank you for watching or listening check Che out all episodes of Channel Insider partner POV on Channel insider.com or watch on YouTube at channel inssider unor news and Trends you can also listen to the podcast version of every episode on your favorite podcast listening platform like apple podcasts or Spotify don't forget to like subscribe and follow wherever possible so you never miss an episode once again I'm Katie boso and I'll see you next time n

This transcript was generated automatically from the video's captions and may contain errors.