Video: Scaling innovations with AI at the Edge - VSCO’s Real-world App Modernization Journey | Duration: 2080s | Summary: Scaling innovations with AI at the Edge - VSCO’s Real-world App Modernization Journey | Chapters: Introduction and Overview (91.065s), Cloudflare Partnership Journey (212.42s), Modernizing Customer Applications (340.555s), Introducing Canvas Tool (466.59s), CloudFlare Integration Advantages (594.90497s), CloudFlare's Key Benefits (773.25s), Rapid Development Benefits (1063.2201s), VSCO Platform Overview (1289.89s), Modernizing Infrastructure Benefits (1413.5249s), Future AI Development (1637.135s), Future Innovations and Advice (1707.99s)
Transcript for "Scaling innovations with AI at the Edge - VSCO’s Real-world App Modernization Journey": Hello, everybody. Thank you for joining us. I'm Sai Chaudhley from CloudFlare, and we have a special guest here, from VSCO CTO, Chris Hare. Chris, if you wouldn't mind introducing yourself. Sure. Absolutely. Hey, everybody. I'm Chris Hare, the CTO of VSCO. And, yeah, I've been with Fisco for two and a half years, leading up the technology team there. I also run pro program management and IT for the for the company as well. Awesome. Thank you, Chris, for joining us. So folks, today, we're gonna be talking to you about scaling innovations with AI at the edge, and we're gonna really dive into VSCO's real world application modernization journey and learn from their best practices. So, Chris, if you wouldn't mind, just as an introduction, what does VSCO do and, what what is the broader remedy of your role there? For sure. Yeah. So VSCO is a community driven creative platform for photographers and videographers, and our mission is to nurture creativity so you can make it. And making it can mean a lot of different things. So for us, that means going beyond just editing tools to help you make your photography into the final product that you wanna get out into the world, but help you make it as a business. So, we have tools around collaboration and finding work and running your business, including a CRM that's purpose built for their needs. We started out building film inspired presets, and, Photoshop and everything, and then we we branched beyond that in building editing tools. And like I said, now we're evolving to doing so much more to build a platform for everything you need to run your business as a photographer. And as CTO, I'm responsible for overseeing our technical vision, strategy, and execution. And, as I I mentioned before, I I also lead up our program management and IT support across the company as well. I I love it because, you're a unique individual who has a view of strategy and how it is executed. And it will we will have a lot of discussion around some of that. And, I think that will provide a lot of richness to some of the, listeners and viewers. Okay. So let's start at kind of the start of your journey. How did you hear of Cloudflare and and what was that experience like? Where did you start with Cloudflare? Yeah. For sure. So, I mean, I over the years, I've been very familiar with Cloudflare, all the way back to its beginning. And, knowing it very early on became a leader in the space, for content delivery and security. And, you know, for VSCO, the first time I connected with CloudFlare was back at the beginning of my tenure attending AWS Reevent and, you know, meeting up and talking about some of the challenges we had. Some of those were amongst storage and networking, and we were looking for better ways to store images we had on the platform and, secure our platform for our creators to make sure we didn't have bots or others scraping images. We initially focused on the layer seven solution and, NR two for ways that we could move, some of our Internet traffic as well as our storage over to Cloudflare because we felt like they would be a good fit. But we've really found that there's so much more that Cloudflare can offer us, particularly as we we get to later phases where we've introduced generative AI into our product, using the workers' AI solution. And our experience has been that CloudFlare has been a great partner, highly collaborative, working with us, listening to our problems, you know, looking for best fits for how they can help be part of our our solution. So it's really become a strategic strategic partner around us to help us modernize our decade old technology and build for the future with AI. I love it. And you have quite a few nuggets in there that I want I want to emphasize, a bit more. So one of the first things that you mentioned was the you've talked actually about your journey with CloudFlare from the start to end or or to the current place you're at. And you talked about how you started both with the connectivity side and the security side, and that's something so important to so many of our customers who are fast moving technology firms or fast moving in their industry and leveraging technology to serve their customers. But the other part is how you have expanded to the other security use cases and then using our developer platform to build new capabilities for your customers and users. And now as the generative AI craze and interest is so high amongst your own customers, you're leveraging Cloudflare to provide the primitives to help you integrate generative AI capabilities into your current applications and new applications. I think that is something that is so interesting about application modernization. We often think of it as just legacy applications that are back end. But what you're talking to us about is it's also about your customer facing applications that might have grown older over the years that you'll have already been in been thriving as an organization. And and it is about building new applications, especially in the generative AI world where previously there wasn't a prior. They're coming up with something brand new, in this space. So this is this is awesome. So I I wanna get into that, the the new applications that you all you all are building. We know one of them that has made the airwaves, as Canvas. But can you tell us more about what is Canvas? Why was it required for your business? And, we can get into it, we can get into the kind of key moments as as we go through this. Sure. Yeah. So Canvas is a new tool that we've launched this year, earlier earlier in 2025. It's an AI powered collaborative move forward for photographers, and it it it it's designed to help you out in the ideation phase where creators and clients are are need to quickly iterate on visual concepts. So we like to say it helps you bring your ideas to life faster. And, the the imp the the impetus of this was really looking at how photographers are already doing mood boarding in their process. They do a lot of this offline. And getting together with clients offline to do this is really hard and really expensive to do. Oftentimes, you're not even in the same state or country. And so, so sometimes it's even foregone because it's not something that is practical to do. And we think that technology enables us to build a better option for people these days and make it digital first. And AI is is also, a great technology to leverage in building a product that helps bring ideas to life really quickly. You know, as as generative AI shows, you can just take a simple prompt and you can generate images and everything from it to get ideas out there. So generative AI became this this awesome opportunity to not only create an online version of it, but bring those ideas to life much faster. We felt like this was the the right mix of something that our photographers, our creators are looking for, to to to solve a real pain point in their overall business, process. Awesome. Awesome. I think the the importance of how you've identified what is the journey for your own customers, just like we are in the business of understanding the customer journey for folks like y'all, to deliver the best experiences for your customers. I love that journey there. Let's get into kind of the key moments a bit more tactical in nature, around how you decided a new application was necessary and and, how Canvas has, then come to be? Yeah. So, I mean, we didn't really have any tool like it, any any surface like it in our platform yet. So that was really where it, you know, it's clear. We needed to build something brand new. And, you know, in a lot of ways, it it has similarities in product experience to something like Figma or other infinite canvas sort of products on the market. What we needed to do is also create real time communication between customers. And, of course, we talked about the generative AI piece, which is a real cornerstone we felt to to the overall process. So, when it came to to to running our generative AI, models, for for the product, we really we we we essentially did a bake off, and we looked at all the different providers and and companies that could help us power that. And CloudFlare was amongst those given that we had already started working on the layer seven and r two solution finding great results. And, ultimately, CloudFlare won that bake off, workers.ai was the best choice for us when it came to building the the most performing experience for for our customers and getting us the the the flexibility for our dev development team to, choose models that we wanted to work with, modify those models, deploy those models really quickly, and and, you know, be able to do what we needed to do to build the best experience for our customers. The final thing is I stated x factor for us was the combination of r two and Workers' AI. The ability to be able to get data in and out of CloudFlare through R 2 working with Workers' AI is critical to the success of the product that we built. I I love it. This is something that, there this is so much to enjoy, and we could have dug in for many, many hours on each of these items. But I did wanna emphasize the piece around workers' AI and R 2 as a, a place for storage of some of the the, the entrance entrancing data. One of the things that we have found with some of our customers is many of them are already building tools and apps in public cloud, spaces using public cloud providers. And we understand that there is a egress cost that comes with trying to take in data in and out of those, clouds. And so for certain use cases and where you have a lot of egress happening from those cloud providers to reduce those costs, that's where, a solution like our, r two and using it on top of our workers, workers' AI, has been so very powerful. So I I love that use case that you bring out there. Okay. So we've talked about the challenge and how you've built out our Canvas. Let's get a bit more, kind of technical in that sense around, the architecture around it. So let's start just with the key use cases for the for why CloudFlare was so useful in this context. What were what were those requirements that you were trying to, kind of solve for? Sure. Yeah. The the key requirements to us were multifold. You know, first, we want to make sure that we're building perform for for performing experience for our customers. We want them to be able to interact with a web based, browser based, infinite canvas, start typing in a prompt to generate to generate an image of an idea that they have, and have it return results in near real time. So that was one key, aspect. From a development team standpoint, we wanted the flexibility to bring the right model to the, place where we were hosting the inference. And we needed to be able to modify the model, not just work with a generic open model. So in this instance, we use Fluxnel in order to power the experience. And, we modified it in order to do region based prompting. And, we so we needed that flexibility. We needed something that was scalable. We didn't know what the demand of of our product canvas would be out of the gate. We knew it would probably be a product that's gonna grow in usage over time, and we need to have infrastructure that can help us scale and grow to demand. So we're we're being efficient and effective with the cost as well as we're able to grow and scale for tomorrow, and deal with spikes as they come up too. And then ultimately, the cost of efficiency is really important thing too. Inference is expensive. So we need to, again, have a flexible model that allows us to do this at a reasonable cost for the scale that we have today and for tomorrow. So it needs to be sustainable. That that that makes sense. And then yeah. Go ahead. I I was just gonna say, I think that that makes so so much sense because we keep hearing about, you know, the benefits of of using AI in your applications. But there's a very real cost to both the training, if you're engaging in that, as well as the inferencing, especially in a mood board where, you know, folks can kind of brainstorm and and test out different ideas. There's a constantly new request coming in. And so making sure that it is valuable for your users and practical for your business, is something that we've realized is so very critical for, technology forward organizations like yourselves. That's right. To that point, I guess, where did, kind of, how how did you find CloudFlare was was really useful in that context? Yeah. So, CloudFlare was really useful because, it really met all the criteria the best. When it came to latency, it it was the best when we did our bake off amongst, you know, several different providers. When it came to cost efficiency, again, it's, it's kind of a no brainer, especially when you marry it with r two in the the zero egress fees. You know, images and generating new images, getting them out, bandwidth is going to be expensive, storage is gonna be expensive, and inference is gonna be be potentially expensive. So, you know, you're looking across all three of those and seeing, like, how it all adds up in the total cost of ownership. CloudFlare was no brainer winner as well there. And the scalability, you know, I have high confidence with workers.ai that we can scale, to the workload that we need. So, we have done things like banking specific amount of inference time that we're planning on, but we have the ability to spike out beyond that if if necessary. And without having to do anything like, you know, spinning up new new instances or, you know, is spot available? It's, no, it's it's it's serverless and and and works works without the complication of having to maintain and watch it to be able to scale. And the flexibility, finally, like, the the ability to bring our own model, to modify it, to to to adjust it to our needs and then have that be able to be deployed so we could use it on infrastructure was, a must have for us, and Cloudflare checked the box there as well. I love it. One one of the things that I'm hearing, you talk about is Cloudflare fits into the way that you work, and you wanted to run the models and also the inferencing and make it make sure that it is closest to the users without your developers becoming experts at your infrastructure. And so now your ex your developers are essentially able to focus on accelerating the road map, building new capabilities, dealing with any of the feedback that customers provide to you. Yep. That's right. And, we we think of this as what we think of this as a core tenant of our connectivity cloud, at about kind of simplifying the routing, enabling you to get more observability and and offloading your infra, infrastructure onto onto us. That as you call it, the serverless cape capacity here. Okay. So the last part, we've touched on a bit of the benefit side. But would there be any metrics that you're tracking or the way that you're thinking about the success of these, of this program? And I I I want to mention that you obviously have app modernization, across the board. It's not just with your latest application, but you're also you've touched upon protecting your entire suite with with Cloudflare. So how do you how do you think about, these kind of, benefits? We'll we'll touch upon the rest of the, side of things, but just for Canvas itself. Yeah. I mean, Canvas, the first thing was the the fact the the ability to be able to come to market really quickly with an idea and bring it to life and give it as another part of our platform to our customers, the the speed at which we were able to do it by using CloudFlare was, tremendously fast. From the moment that we started deploying it out on, in into production, building it for production and getting it out to customers. It only took us a few months. And and I know we wouldn't have been able to go that fast without ThoughtFlare. And we've been able to continue to to develop really quickly new ideas and new capabilities for the product. So not only is Canvas bringing the opportunity for rapid iteration for our creators with their clients, it's it's creating this rapid iteration that we were able to have with our with our creators, and and bringing things to life. And, you know, so ultimately, it's it's delivered on the promise for our creators. We have introduced a new CRM tool and and workspace in this summer, which allows creators to run their business and manage their relationships directly with clients. I think the future for us is, finding ways for us to to seamlessly integrate tools like Canvas and Workspace together so you can launch your mood boards and start that that process directly with your clients when you have a specific project you're working on with them. Ultimately, like, CloudFlare has really helped us also have great predictability in our cost and efficiency here. So without the egress fees, without overbuilding infrastructure, it's been a huge win for us to be able to, have a new product come to life without it costing costing us more than we we would potentially make as a business off of off of the tool. So we we we don't have to worry about overspending or rapidly rising costs that, could put us out of business, and and make the product nonviable. So that's been a huge win for us as well. Awesome. So you you've touched upon, Workspace. So let's get into that, a bit more. And this is beyond the beyond your, the the latest and greatest canvas, but to your kind of existing platform. Can you speak about what your platform looks like, and what what has been the value of kind of using Cloudflare to protect and connect those existing platforms that you already had in your business? Yeah. Absolutely. Yeah. I mean, our platform is so much more than just Canvas. It, it starts with our mobile apps. We have a core VSCO mobile app. We launched a new app this summer as well called, Capture, which is a, kind of an alternative camera for your phone that allows you to, you know, take the magic of our presets and bring it right into your camera so you can select presets before you even take your shot. And it's got tools in there that are built for photographers, like helping you kind of divide up the scene and see, like, the rules of three. I'm definitely speaking, like, some photography lingo here a bit. But, so we've got we've got our mobile apps that are a big part of, big part of our platform. We've got our core website that includes editing tools for the web as well. We've got Canvas, like we talked about before. We have a CRM that helps you run your business called Workspace that we recently launched. So we have a a large suite of tools, and underpinning all of that is a common platform where you have one universal login for VSCO, and you're able to carry, along workflows amongst in between all these different applications. And, so there's a lot that's powering that. We've, you know, we've got our own API and infrastructure as well as, you know, data storage, file storage, images that we're storing. So there's there's a lot here that we've built over the years and that we continue to add to the platform. Awesome. We think of that as and I think, you know, it's it's broadly, thought of as for your existing apps, rehosting the infrastructure and replatforming it so that it provides you the single platform, is very critical. And I I I know you have some thoughts on, some of the benefits out of that. Can you speak to that a bit more? Yeah. So, you know, where we started working with Cloudflare was looking at how we modernize our current infrastructure that already existed. When we when I when I started two and a half years ago, the amount of images that we had from all the posts that people had done at VSCO was was over 13 petabytes, a tremendous amount of of amazing images that people had shared on the platform. And, it it had grown really large on another plat on another cloud provider. It become a a a real source spot as far as, how we continue to to host that that amount of data at scale, and, you know, be able to still have a performant experience for being able to pull down those images and view them for for VSCO creators and their audience. The the cost was getting out of control, and so we had to look for ways that we could modernize that and make it better. We also wanted to up our ability to be able to secure that data because our images that our creators bring to our platform are are really like, we we treat those really as as really important content, as treasures, if you will. And we we we think it's the best thing to protect that that, amazing work that our creators have built for our creators. So, those were important things that we needed to, to try to solve and improve. Overall, by going to kind of a multi platform approach and and moving those images over to CloudFlare, we were able to save over a million dollars a year after we enter we migrated. We we did all that work not only to get cost savings, like I said, for security. And by leveraging WAF and the m advanced bot protection tools that Cloudflare has, we were better able to protect the ability for, image those images being scraped by bots or other bad actors on the Internet. Performance wise, we saw huge improvements. We always wanted to put our API in front of layer seven so that we could we could speed up the the performance that we get in our general API for our for our product. And using CloudFlare and using Argo, we are able to to create a sick, six x faster experience for some of our customers, particularly in Asia and South America where we have tons of global customers, interacting with the platform every day. And and finally, we are able to simplify our platform because we have built some technology to do things that Kotler did for us. When we started investigating whether we should come over for layer seven and r two, we we we uncovered the the image service that, the that Cloudflare has allowed us to replace our own home built image resizing service. So it greatly simplified the overhead that we have as a as a company when we're operating less things and we could offload that to Cloudflare. And we we save money as well, by using CloudFlare service versus building using our own homegrown solution. So there's a lot packed in there, but there was a lot of benefits that we got through modernizing our platform, partnering with you all. Thank you. Thank you for trusting us. And just to recap that it was cost savings across just the pre platforming with CloudFlip, performance improvements, security benefits. Especially now, you're y'all are almost, leading in this area where we've seen just a huge amount of organizations, creators, worried about what is gonna happen to their creations as content, images, media, etcetera, as AI crawlers scrape all of that, information. They want back control it, where we think of it as a content independence. That's what they're looking for. So looking ahead now, we've had we've talked about your new apps, your existing apps, and how you're protecting and connecting them and improving performance. Looking ahead, and we're running out of time, so want want to, definitely get to this spot. What what's the next step that you're thinking about in your, AI and app development journey, and, and where might Cloudflare fit into that? Yeah. The one of the primary places we're looking at, continuing to evolve the platform is on the AI side. AI keeps bringing better capabilities to build amazing products for people every day. We're getting set to launch some AI based editing tools in our core app experience, and we know that CloudFlare is gonna be the provider and partner that's gonna be able to help power the best experience there. So that's top of mind for us. We're always looking to up our security as well. We continue to to tune things with bot protection. We know that CallFlare is constantly working on security and upping things there. Very, interested on some of the the great the the great announcements that are coming out in the developer week. So, we're going to be looking at those and seeing how we can take advantage of them to continue to up our security too. So, yeah, a lot of a lot of great things on the horizon and and, you know, love that we've got a partnership that can really help us build for the future. Yeah. Thank you. And, VSCO fans who are listening to this, Chris just gave you a sneak peek that there's something cool coming out, so watch out. Watch the airwaves. And absolutely happy to, to be a partner and and a supporter of of your initiatives here. Last part is what would be kind of, for those listening in, what would be best practices or advice, that you would give folks when they're thinking of modernizing their apps and infrastructure? How how would how would you suggest those? Yeah. I mean, you know, just look for the I guess, the the the overall tenant and principle I always use personally is where's the highest leverage for the business? And, you know, with infrastructure, there's just so many different options that you can use these days, and that and, you know, it's it's very possible to be in a multi platform world or to replatform when it when it's the right thing to do. So, you know, it just starts with looking at your business and seeing where there's, you know, big opportunity and looking at, like, what options are on the table that you could consider to take advantage of opportunity. So with with with CloudFlare, I think, again, I I I wanna emphasize how, workers AI are to the new, you know, data connection, technology and announcements that you have coming out. The the marriage of data and AI is really important to build the best experience, And there's may there's some pretty good synergies going on with your network, with your data storage story, and with workers' AI that makes a lot of sense for the type of use cases people are building with for generative AI products today. So to me, that that's, like, a clear reason why the consider Cloudflare as, you know, something that could really power that stuff for your business. Thank you. And as, we've been saying, we we love partnering with you. And for everybody else listening in, do check out our birthday week, that is ongoing right now. We're releasing a lot of capabilities, that can be used in a lot of organizations today. And as Chris mentioned, we have our innovation weeks like birthday week right now as well as, developer week in, next year, where we'll be talking about many more, generative AI capabilities as well. With that, Chris, thank you, for joining us. Thank you for enlightening folks on how Canvas has been built, Workspace, and a lot of your existing mobile and web web applications today. Thank you for your time again. Thanks, Sai. It's it's been great partnering with you all. I'm happy to happy to to share some of our story today with everybody else. So thanks. Thank you.