Enterprise AI Platforms

Today's show is sponsored by Vasion Print. Technology should drive your business, not slow it down. But legacy print systems are a resource suck. Don't waste time and money in managing print when there's a solution. Vasion Print, formerly known as Printrologic. Eliminate resource-intensive print servers and make the switch to an intuitive, cloud-native solution that makes printing effortless for IT teams and end users. Reduce the tax surfaces and secure your environment while enjoying cloud-native flexibility that's easily supported in a hybrid workforce. It's not a patch or a fix, it's a truly seamless solution for print. See for yourself by scheduling a demo at Vasion.com. That's V-A-S-I-O-N.com. Today's show is sponsored by DoIt. CloudSpend is just the surface. What really matters is how it drives performance, reliability, and speed to market. DoIt Cloud Intelligence aligns every dollar to the intent of the workloads. DoIt is trusted by companies like Fiverr, HackerRank, Current, and SuperBet. DoIt helps teams go beyond dashboards, surfacing hidden inefficiencies, resolving incidents faster, and aligning cloud operations with business priorities. One platform, unified signals, better outcomes across engineering and finance. Visit DoIt.com. That's doit.com to unlock intent-aware finops at scale. DoIt Cloud Intelligence. Cloudcast Media presents from the massive studios in Raleigh, North Carolina. This is The Cloudcast with Aaron Delb and Brian Graceley, bringing you the best of cloud computing from around the world. Good morning. Good evening, WebR. And welcome back to The Cloudcast. We're coming to you live from our massive cloudcast studios here in Raleigh, North Carolina. And it is just Aaron this week, but we have a great topic and great guest for you all. We're going to be talking a little bit about AI enterprise platforms and kind of dabble into AI security a bit as well. And for that, we have Shai Levy, co-founder and CEO at Unframe. And Larissa Schneider, co-founder and CEO at Unframe as well. So Shai and Larissa, welcome to the show and give everyone a brief introduction, please. And a little bit about your background. Hey, hey, great to be here. I'll go first and I'll pass it on to Larissa. So I'm Shai, I'm the co-founder and CEO of Unframe. Heavenly focused on AI. We're going to talk all about that in a bit. But before that, I was the co-founder and CEO of a company called No Name Security, which was leading the API security market in the cyber security space for about four years until our acquisition by Akamai for half a billion dollars grew that company all the way from zero to about 250 people. It was very successful and happy ride. Before that, I had another company profitable from day one with no investors. Before that, I worked for Facebook. And I started my professional career in the intelligence unit and the IDF where I served four years in various cyber security roles, software engineering roles. So basically, I've been doing cyber security, software engineering and AI for over a decade now. Fantastic. Hi, everyone. Thanks for having us. I'm Larissa. One of the co-founders and CEO of Unframe. China and our third co-founder, Adi, we met during my time at No Name. I've always had a background on the go-to-market, marketing partnership side of things. Before that, I spent six years with so at Nutanix, someone like the Cloud Management Infrastructure side of business, quicks in cyber. And now I'm really excited about AI and what we're doing here. Fantastic. Yeah. And yeah, full disclosure. Larissa and I spent some time together at Nutanix. And we had definitely some fun rides and some interesting experiences there for sure. Stories we could tell overbears. So let's kind of jump right in. So like, as I mentioned, today we're talking about a little bit of AI security and a little bit about enterprise platforms as well. And so let's kind of start there with what are the problems or issues that you're really seeing with AI development today? Because we talk about it a lot on the show of like, hey, this Gen AI thing, it comes along. And then just like anything else, you're kind of-- even though AI is very mature, AI development and especially around LLMs is a little bit nascent, right? And it's kind of early days. And it has its growing pains just like anything else. Larissa, I want to go first. Absolutely. So I think what we're often seeing is a lot of enterprises are-- and we're heavily focused on working with enterprises. We see that they're trying a lot of things. They have a lot of ideas. But they're often really trying to reinvent the wheel in many aspects, like trying with your projects, building something internally, spending month at end, organizing data, mapping data, transferring data, you know? And like, doing experiments on training the own model. And so on. So often, they don't really actually get to show the value. Or if they do, it takes them month at end, just getting something out ridiculously expensive. And we really felt that there needs to be something that's much faster, but still gives them like really good business value that they can see very, very quickly. And so I think that's the problem that we jumped on and started working towards. Yeah. If we zoom out a bit about it a bit. So I think AI came into the world, and everybody started thinking like, you know, how the future can take us. And they started talking about AI replacing humans. And like, all of-- you know, it's something very much in the thinking of everyone. But in the practicality, when you try to build something with it that is actually useful, that still requires, you know, thinking about the problem you're trying to solve. That still requires a certain dedication from the development team around that problem. You know, I see a lot of companies that their business is around finance, or their businesses around cars, mobility, you know, real estate, whatever it might be. And suddenly, they have internal teams trying to create, you know, AI systems for ticket management, or AI system for sales management. And this makes no sense. It's not their core business. It's true that AI makes developers more efficient. But do you really want your developers to spend their time building an IT management system? That's not your business. And you don't do it. Like, you have service now. And you have Gira. You never thought about doing that in the past. Well, why would you suddenly try to build everything in house? And I think, you know, and I think a lot of companies are realizing that that approach of will build everything internally, and will have like our own team build a bunch of AI use cases actually doesn't prove itself. Because it's very easy to build something that is like 50% right. But no one wants to use something that is 50% right. Like, no one. There's 0% adoption to those. And so they invest a lot of time in AI, but they don't really make a bang with it. And then it's really frustrating to a lot of enterprises. And on the flip side, though, they always know that the enterprises that will adopt AI and will have the right strategy, or will be able to harness the AI power, will significantly beat their competitors. You always know that as an enterprise. And so this is what they fear about. And to us, obviously, this is why Unframed came to light. Enterprises that choose to work with us do enjoy that benefit. Because you're not devoting your own R&D team that is focused on your main business to solve problems or areas where you can harness AI to make your org better, to make your enterprise better, to make your products better. I really think that companies that work with Unfram have a competitive advantage, honestly. Like, that's how I view it. I like that. I like that. And kind of a follow up to that. It reminds me of the early days of cloud, if you will, in a couple different ways. You have these silos of everyone wants to play and everyone wants to kind of go do this project. And so you have this fragmentation of somebody's doing something over here and somebody's doing something over there. And you don't have this efficiency of consolidating all of those workflows into something centrally. And then you have this idea too of like, oh, the shadow IT days, right? Well, now you almost have shadow AI days, right? And so in my mind, this is where this AI platform, the term, kind of starts to come into play and starts to make sense. And I'm seeing this more and more here recently. But I'm wondering too, because it is so early on in this, what does it mean to be a platform? Like what is your definition of a platform? And what does it look like? And what are the advantages of a platform? I think that's a good question. And I think it's important to make a distinction here, right? Because the word platform is heavily used in tech in general. You can be a single product and you call yourself a platform. You can be something that is made for developers and you call yourself a platform. You can be something that is hosting a lot of different applications, not necessarily for build, and you would call yourself a platform. And so almost everything is a platform. So it's hard to define. So I don't blame you for saying that it's kind of hard to put your finger on it. I think there's few angles to AI platform. Some approach developers. They say, let's make developers-- make it easier for them to build. So the internal developers-- inside the ore can code in a better way, in a more centralized way, in a more governed way, things of that nature. There's this approach of a platform. On the flip side, there is very like point solutions that are calling themselves a platform. They're not necessarily, you know, they are only in a single ecosystem. They're a product for one use case, but for the benefit of it, for the benefit of the name of sounding bigger, being able to get a larger ticket, ticket size, they call themselves a platform. So this is the confusing part. The third angle, which is what untrame is really focused on, is somewhat of a delivery platform. We call untrame like a managed AI delivery platform. The idea is that you get the benefits of all the different solutions that you need, aggregated, easily governed, managed long term, but you don't need your developers to create every single AI use case. Like that is the approach of untrame. We are the ones that are creating the different solutions that you have available on this one platform. We are using the same building blocks and the same reusable components to assemble these solutions for the various use cases that you have, but we are the ones maintaining it. We are the ones making sure that you adopted, that you enjoy it long term, that you get more features, things of that nature. So this is our approach. It's like AI delivery platform, not just like something for developers to code on. There is a few, if you want to add something that I forgot, I mean, feel free. Now I think it's hit it perfectly. As you say in the so fast evolving markets, there's a new platform popping up all the time. And I think they all have different reasons for being. We just see that most of the time on the enterprise level, the requirements are very different than some of the more B2C and developer focus platforms. And that's why I think ours is somewhat different than a lot of the AI platforms that you see right now popping up. >> I like that. Thank you. Thank you. And, Shai, let me kind of ask you specifically going back to kind of no name and API security days, right? Like considering your background in APIs and API security, how does that knowledge transfer forward into aIs? Because obviously you're seeing lots of AIs being API driven and you're kind of seeing like open AI is oftentimes becoming the API standard, right? You're seeing everything being advertised as like open AI compatible. And like tell everyone when it comes to APIs for AI and security, like what do you think and how do you think about this space? >> Right. So I split it into two. There's the security aspect and there's the API aspect. I think in security with security goggles on before this AI revolution, for a very long period of time, there was not something very like important to chase, some great innovation to chase. And so I think security, you know, a lot of organizations started focusing very heavily on security. Let's secure what we currently have. Let's make sure we don't break what we currently have. Everything was around that. Now AI is different because this is a revolution that can make or break your company. As an enterprise, as a large organization, you don't, you no longer have the luxury of just saying, let's stay secure and make sure what we have is working. This will not work because your competitors will adopt AI. So trying to hold back on this innovation can really break your company. And I think most enterprises understand that. So they rush to adopt AI. So from a security lens, you have to kind of secure on their, as they go, you can't tell them to stop because if you tell them to stop, you risk the business. So you have a lot of challenges of, okay, how do I keep them secure, but I don't hold them back from adopting. I heard initially of organizations saying, we, you know, we don't allow anyone in the org to do chat GPT or to do, you know, those. This is, to me, this is a very bad idea because the competitors that might be more lenient are actually allowing this. So you might be more, you know, govern, secure, whatever. But if you have no business left at the end of the day, then you kind of, you harmed it in a different angle. So to me, the security goggles that you need to have today is let's, you know, let organizations adopt AI from the security perspective. But let's keep them safe wherever we can. So let's think of how we govern them in a reasonable way, not by holding them back, letting them flourish. But in this all has to happen very fast. And so I think our advantage in on frame, coming from security background, I know how the good needs to look like. I already know what good needs to look like. So when we created the, the AI delivery plug, when we created on frame, when we create solutions, I already make sure that what we deliver out there is, is, is the good standard because I know even if I do something medium standard, probably organizations would still adopt it, but you know, I don't want to be in that, that position because I know they, they rush to adopt AI. So I think that this is the security angle on the API angle. I think organization started to really understand it when MCP came out. If you think what MCP is, basically it's API calls, right? There isn't, there isn't much there. Like it's, it's a protocol for performing those API calls, but those are API calls. And immediately my mind started going on, how do you secure those MCP calls? How do you make sure you don't, you know, there's no malicious actions going on? And I'm sure there will be like, you know, the existing DLP's or AI security companies that are dedicated for LLMs are going to do some work around that. But to me, the angle that is most interesting and the most, that I like to see the most is that finally, I feel like there's a revolution that you cannot just say, oh, it's unsafe. It's weight. No, no enterprise is going to wait. And so it's, it's fun to see finally some fire going on in the security orgs where they're like, okay, let's go. Like we need to, we need the org needs to adopt the I let's figure out what's the best way to do it. And I think it's, it's fun. That's how I see it. Yeah. No, absolutely. And given the typical organization's enterprise, the security posture is like, oh, you know, we want to change as little as possible because change almost represents risk. I mean, that makes perfect sense of like, hey, we finally have this moment. And to propel the security industry forward a little bit, right? And, and, and so it got, Lerse, I kind of want to flip there and go, okay, with your, you're kind of background in operations and, and, and talking to enterprises, where do you see like these AI efficiencies that I was talking about earlier, right? Like in my mind, there's this AI development lifecycle and there's life cycle management. And you've got AI models and you've got AI data sets and you have to match everything up. But how, how does the enterprise and enterprise AI, how do they think about what is operative good operations look like? Yeah. So at, at, I'm from in general, what we do is we always start from the use case. That's what we focus on. We really want to make sure that the solution that we're bringing to the customer is the most tailored, most suitable exactly for what they're trying to achieve because we believe that's what you can currently get most accurate responses with and that that's what we'll get you the best ROI at this stage. So we, we always start there and in most cases, what we've really noticed over the last year or so is that pressure from the top down. So the boards and every board meeting around the world in the last year, it's been asking the, the sea level leadership, what is our AI strategy? What are we doing here? And so like the, they really have figured out this like adopt AI of all behind comparing them to competitors. And I would say most of the time where we're seeing the impact, like we really try to start with that highest business impact type of solution first. They either trying to replace a lot of the human, very, very labor intensive workflows in return, increase efficiencies, of course, reduce the human error side of things. But also often we see use cases around there being efficiencies that they're trying to make sometimes either replacing people or just getting a lot more mileage out of the teams and resources they have in house. So I would say those are probably the use cases that have highest impact right now. But if you're talking about operational efficiencies as a whole, as an entrepreneur, I think that's a very important point for us from a platform perspective as well. Because we're really, I think building our whole team, our whole company, our whole product on top of like an very operational efficient mindset as well. Because every single building block that makes up our platform is something that we came up with together with real customer projects. We've been building them in the most generic way possible so that we can reuse them hundreds of times over and make them very relevant for every customer environment. So we try to really take this moment of like having build a company at the height of kind of the AI buzz, if you will, but this really the turning moment and try to make everything internally as well as externally with customers the most efficient possible I would say. Fantastic. Fantastic. So let's go ahead and because we've been talking about Unfray, but we haven't really dug into the details of the platform right. So let's talk about that a little bit. So give everyone an overview. Is this a SaaS service or where does it kind of fit in an AI development stack? Because I think we're at a place in the industry where the tools, like everyone's kind of using a bunch of things cobbled together and they're, again, going back to kind of inefficiencies and fragmentation. So give everyone an idea of like, what does this look like maybe before and after? Larissa, me, you? You. All right. Well, this is about the second we're going to cut out. So, Unfring is a managed AI delivery platform. The problem we identified in the market is that while adopting AI should be a superpower, but instead it's a struggle. Generic tools don't really work or don't fit your use case and something that has 70% accuracy, has 0% adoption. In-house development takes very long and requires expensive talent. Point solutions don't scale as you have a lot of use cases and a lot of things you want to cover. And traditional consulting is slow and costly. And so what if you could just say what you need and get it, you know, as simple as that? And this is basically what the Unfring managed AI delivery platform is all about. It's a new way to get AI native enterprise solutions. You describe your use case in the old fashioned way of sitting and talking about it with us for about 15 minutes. What are the relevant integrations and we can integrate into everything? What are the AI capabilities that you expect? What do you want AI to do for you? And that's about it and like how a good result looks like. We meet again three days later and you already have a production ready enterprise grade solution ready for you to try the actual solution that you ask for that you describe. You usually try it initially on dummy data and if you're impressed, you can go ahead and integrate it with the same integrations that you said are relevant. And actually go ahead and use it. And Unfring operates in a complete risk-free model in the sense that we only move to licensing if you're happy with the outcome. If you say yes, I get the business value, only then we move to licensing and we charge per solution per year. And the reason this is a platform is because you can do it on many different use cases. You know, you initially start with one use case with us and then you say, oh, I have another. I have another. I have another. The reason we are able to do it, turn around that quickly, being able to launch so many different solutions for, you know, on your environment, which is basically your AI platform, is the way it operates. So what we have behind the magic is hundreds and hundreds of deep technical building blocks that make up our platform that can be tailored or moldable for the specific use case that you need. We do that. We do the tailoring or the configuration of those building blocks with a layer we call a blueprint. You can think of a blueprint like a small little spec file that someone just, you know, assembles together. It takes about two days to do it, which is what we do once we talk about your use case. That configures the different building blocks for your use case and gives you a turnkey solution ready for you to use. And those building blocks because they're of high quality, because they have seen a lot of different use cases being built on them because they keep evolving over time, because this is what makes up our platform. The solutions you get are of high quality are tailored specifically for a use case. They're not generic. They solve your problem. You know, people can just log in and use them. Now, what needs to set and code and configure or like do complicated stuff. And this is why I said that enterprises that work with on frame really get to harness the power of AI. You know that you talk to us about a use case today or three use cases today and two weeks later, you have them in production. It makes a huge impact. I'm telling you, the ROI that companies see from working with us is incredible. And Larissa, maybe you want to add something on it. Yeah, please. If I try that back to how we started this conversation and kind of getting enterprise value quickly, like the way that they work with us, you don't have to be a developer, you don't have to be an analyst. You can be the person that is working on your legal team, managing a lot of contracts. You can be the financial analyst working on your income statements. And you have a problem that you think AI could help you with and like modern AI native software. You could just work with on frame and get your solution pretty quickly. You don't have to wait for an internal developer to become available to solve your problems. And you also don't have to find hundreds of thousands of dollars upfront to invest in custom software through a consultancy agreement. So really you get to see what would the solution look like. You can try on frame and only after that will have the proper budget discussion. And that's for us the best way we found how to show impact, show value of what AI can do for a business on enterprise level. And I think really it is enterprise grade. Like, you know, the security's perspective, it's all enterprise grade, the auditability, the security settings, everything in it. You know, we support deploying on the customer environment. You know, this is a cloud podcast. They can deploy it on their own private cloud. They don't necessarily have to use it as a SaaS. So everything stays secure in their environment. So it really enables them to another barrier that would have been in a different route is lifted when you when you use on frame. Love it. Love it. Fantastic. Well, I think that's going to do it for this week. Thank you both. Larissa, if listeners want to get started and kind of learn more, what's the best way they can get started? Absolutely. Reach out to either one of us. Larissa, I'm from Doddy. I website. I'm from Doddy. I would love to have a conversation. Learn more about your use cases. Happy to put together a demo environment for you and your specific needs. And so let's get talking. Fantastic. Well, thank you both for your time this week. Really appreciate it. And everyone out there, thank you very much for listening. If you enjoy the podcast, please, tell a friend or wherever you get your podcast. If you can leave a review, please leave a review as well. And as always, we're looking for guest ideas and feedback show at thecloudcast.net. Thank you everyone for your time. And we will talk to everyone next week. Thank you for listening to the cloudcast. Please visit thecloudcast.net to find more shows, show notes, videos and everything social media.