Rethinking Legacy Data Infrastructure with Eon Co-Founders Ofir Ehrlich and Gonen Stein

0:00Up until now, the concerns came from human threats. What we're seeing now on steroids is that the same type of threat is coming from non-human actors, agents that essentially have legitimate access to the environment with legitimate permissions. Fortunately for us, it's a very similar methodology in terms of detecting that and protecting against that. But the velocity of that happening is extreme. Think of the non-technical people. They're not even aware for things like security or compliance or who is going to use this data. Maybe their agent that they are building are using other agents and they're not technical to even understand what it means.

0:35It creates a complete set of actors inside the organization, not bound by the rules of the organization and not necessarily running within the premises of the organization, but handling sensitive data. It's a good thing and bad thing that everyone inside the organization can become builders. We live in very interesting times.

0:58Today I know Pryors were joined by Ophir Ehrlich and Gunnstein, the co-founders of Eon. Eon is a cloud backup disaster recovery centric services designed for the AI era. In this discussion, we talk about data, AI, why Google bought out the data of Spirit Airlines out of bankruptcy, and what it means to really manage and use data infrastructure in the AI era. Afiyan Gunan, thank you so much for joining me on the prayers today. It's great to see you. Absolutely. Thanks for having us. Yeah. So one thing that you guys are doing at Eon is, or actually, why don't you give a quick overview of Eon and what it does really quickly?

1:34Because I think that'll set the context for how we think about AI and data and models and fine-tuning models. I think there's a whole stack that's built on top of different types of data sets. And so maybe we can start with what you all do. And then I think we'll kind of walk through how the world is shifting relative to the enterprise data stack. Yeah, sure. So what we do at a high level is we've created a new data foundation that runs in the cloud. And we provide multiple capabilities that allow customers to first map and classify their data across their environment, across multiple hyperscalers and identify what they have, where they have it, what's sensitive, not sensitive, and so on and so forth.

2:11then we provide an ability to easily ingest that data from all these different sources, structured, unstructured data into this data foundation. And the data foundation then provides a very cost effective way of both maintaining the data for protection and recovery, but also makes sense of the data. So it allows customers to very easily access it, query it, search through it, and apply their AI models and LLMs on top of that data that's ingested from a variety of sources. Yeah. And my sense is, I mean, your starting point was really as sort of backup and data recovery and protection service. And I think along the way, you kind of realize if you have all this data from a backup perspective and you have all their customer history over all time, you can start using that for interesting application areas.

2:56What are some of those directions where you're seeing customers take the sort of full history of data that you all have or represent? So as you mentioned, when we started, I said I'm this crazy person starting a non-AI company in the AI world. And the AI tailwind became absolutely insane and made sure that data becomes the most important thing that an organization has. When you can think about it, models, compute, everything is relatively ephemeral, almost zero switching costs. And those are important parts of the infrastructure for the industry. But if you're a company, whether you're a hotel chain, or your food chain technology company doesn't matter.

3:40The most valuable thing that you have is actually your data. And you see more and more companies finding this out. Just two days ago, you saw Google buy something from the bankrupt Spirit Airlines. They didn't buy airplanes. They bought the data. They bought the data for$10 million because they think it's very important. In that perspective, they're using it to train models. I think the rumor too is that the other bidder on the data set was Mercor, right? In terms of the bankruptcy bid process. And so it's interesting, you had multiple different companies in the AI world bidding on a bankrupt airlines enterprise data set, which is fascinating.

4:23Yes. Do you think we'll be seeing a lot more of that in the future? Like, do you think we're going to basically be seeing these like out of bankruptcy data buys? So we've seen it for multiple use cases. That's what's really cool about it. And you see Mercur, you see other companies are continuously already trying to buy data. If you're a tech data CEO today, I can tell that you constantly get questions. Are you willing to sell your data? I hear it all over. And it seems that's going to be a significant trend as you go. I'm hearing about labs going to Wall Street and trying to buy data from hedge funds and try to understand how to map and analyze companies.

5:07So you see data that was accrued throughout the years by companies, which was usually like tapes. It was usually sitting on a shelf collecting dust. And all of a sudden, this becomes very important. And you see companies now realize that first, what I have today that differentiates me than anyone else is my data. And this data is gold. And I can actually leverage that to get more value for my company and to continue building my business when AI is actually coming and fattening the playgrounds. It seems that everyone, even large and small companies, basically have the same playing field. And the only real advantage that the company has today is, of course, the people, but also the data that they've approved.

6:08Because everyone has access to all of those cool new tools. Yeah, it's become a moat. I guess in terms of that, I mean, people have been saying data is a new oil for a long time. And I was always a little bit skeptical of that statement. But I feel like now what's happening is because of post-training and reinforcement learning and, you know, there's companies like Applied Compute and others are starting to provide these sorts of services where you can fine-tune models or open-source models against specific data sets. Like, it seems like people are trying to optimize these things for their own use cases.

6:36I guess in the case of something like Spirit Airline, is it customer support for building, like, an airline app? Like, what do you think they're actually going to do with this information? Is it something else? It's the internal documents? Like, I'm just sort of curious, like what is the reinforcement learning or is it like a customer support agent? Yeah, but I think if you're trying to build agents so they're trying to, you can't just build them a lab, you need to train them on new data, on some training data. And it's very hard to find very good datasets. You see that Hub just released a legal dataset a few days ago.

7:15But you don't find too many good data sets that doesn't look like real synthetic data that can actually be used to really look like the real world. And I think that Spirit Airlines can be used both as an airline company, but also as a large enterprise, as a place where lots of people work, a lot of, you know, the hierarchy, middle management, top management, and workers working together. and you know if you're looking at what other public data sets do you have out there there are a lot of those there's the seriously i'm speaking with companies asking what kind of data do you have what are you training on you here find stuff for example the anyone data is out there in public and people are actually using that as real data from a company because somehow how a company works like.

8:07And the reason is it's so very hard to find data that will help you to work like in the real world. Anytime you see someone building an agent or building a new application, you know, most of them don't really work. You have to go to the world. You have to actually interact with real world companies in order to really build something significant. Now you can do it when you go to customers, so you can buy data and train in-house so when you first release your products, every new product that you have, you don't have to first interact with customers as your initial interaction. So I think you're going to see more and more of that, both by creating new synthetic data in new innovative ways, in addition to getting existing data, whether it's the real data, whether it's somehow mass, when you think about it, it contains sensitive information like PII, financial information, so on and so forth, and actually be able to build real-world stuff on top of that.

9:11Yeah, and Google, obviously, they're in this travel space for a while, right? They want this type of data. They're already monetizing it. This allows them to train it, understand it, monetize it even further. And it's a unique situation, right, that obviously people want to take advantage of, And I think we're going to see more and more of that in such situations. And regardless of that, customers who have existing data want to be able to unlock that existing data as well. What sort of tooling are you all building at Eon to allow people that make use their data for AI applications? Like, how are you thinking about this problem yourselves or what sort of tools are your customers asking for?

9:48So let's go back from the problem statement. Why there are so many tools for data and processing? Why do you need new tools? It's already so many great companies throughout the years and everyone understands data is important. So to put it this way, back in the days, every data team could find their own data, decide what project do they have and get data to do something with that. Very tactical. They were using some great companies, Fifron, DBT, Monte Carlo, all the data tools that exist in order to fulfill their tasks. And for some of the data, they didn't even know existed. It was locked. Why was it locked?

10:32Because there are multiple business unit owners across the same company. And let's say you're a data team leader in some company and you are based in San Francisco, or we are now here in New York. And both are also different business unit leaders. and now there's this thing called AI and even the boss is playing with ShareGPT. So the CEO and the board and the shareholders, they understand that AI is real. So they're coming to you and they tell you a lot. We have a lot of data in the organization. We now realized data is new oil. We can actually activate it with the new tools that we have today.

11:18We couldn't before. Do something with the data. Make it useful. And use AFODAT because it's valuable for us and because it's cool. What can you do? So you say, great, I've done this thing before. I just need to bring to, I know all of those new cool things that are coming out every day in Silicon Valley. I can just leverage them. The problem is, where's the data? So you're coming to us and we are business unit leaders. If you even know us, maybe you don't. But let's say that you somehow got to me. I'm a leader of a business unit. I have data, probably. And somehow you convinced me to give me access to my data.

12:01Now, I don't know what data do I have. I have a lot of people working for me. They have data in multiple systems for the last 20 years. Some of them systems that no one really understands where. that contains production data, that contains sensitive information. You know, there's always this server that no one knows what it's doing, but it's connected to the power that everyone's afraid to turn off because we don't know what's in there. So we have all of that. And let's say that somehow I know what's in there. Now I need to bring engineers and compromise maybe security and compliance and production, uptimes, and to extract the data, just to give it to you and store it in a very inefficient manner.

12:49It's very hard. We understood that there's a problem with how this works because we have different incentives. You were tasked with doing that. I'm tasked with making sure my systems work and I'm tasked with making sure that data is intact, no data is running away. I don't accidentally have the salary of the CEO inside my data, and it's actually going to be used for training or post-training by you. So we at EON solve it in a very different way. We can help you, not me, you, the data team leader, find all the data that's in organization, a very simple way, understand what it is, classify it, map it, understand context layer on top of that, build a semantic layer, and then be able to continuously bring all the data from me that is relevant without compromising production, without compromise security, compliance.

13:48We're actually keeping an audit. And because data is classified, I know that I'm not accidentally going to share with you sensitive information that you shouldn't have eventually in your data. We can do it in a very cost-efficient and performant way. so you can actually do it from all over the place, bring it to you and actually use it. So it sounds like there's three or four things that you're solving for. One is you're aggregating lots of historical and current data for people. Number two is you're able to then mask personally identified information or other fields that they don't want necessarily shared or set permissions on top of that.

14:23And then third is it sounds like all this can then be exposed into AI models for sort of their uses or applications. And the key point, Alad, is that customers already have this data. That's kind of the ironic thing. Customers today already have this data. It's kept in their environment in different forms, but it's locked. It's not accessible. And usually it's very, very expensive, right? So we're able to take what customers already have, convert it into this new data foundation format that's stored much more efficiently, and provide the mapping classification access control and connect it into the AI workflows.

15:00How do you think about security? So there's been a lot of news recently about the labs where they'll have agents, like it's escape sandboxes and do all sorts of things. And, you know, there may be broader things afoot in terms of why that's happening beyond just the agent capabilities, like who knows how these things are set up or configured or, you know, sometimes a little bit uncertain whether, you know, there's that much how people are pushing these things. But, you know, fundamentally, there's a lot of discussion of like AI security. How do you think about that in the context of the enterprise stack, what people should do or not do, how CISA should be thinking about all this?

15:34Yeah. So up until now, the concerns came from human threats, right? So this is not you where customers would come to us and say, hey, we were exposed by this ransomware attack. So during our time at AWS, it was a very large customer that was impacted by ransomware. We thought that they were completely protected using our technology, the disaster recovery service that we managed there. And we learned, unfortunately, that the customer thought that they were protected. They weren't protected because they didn't map and classify and tag their resources properly. So it wasn't protected. And so 60 % of the environment was exposed by ransomware.

16:12And that's one of the reasons why we decided to launch Eon and solve that pain point around human threats such as ransomware. So being able to detect when that happens, look for irregular write patterns and entropy changes and things like that, protect against it, and then also allow customers to recover in a granular fashion and very quickly. What we're seeing now on steroids is that the same type of threat is coming from non-human actors, from AI agents that essentially have legitimate access to the environment with legitimate permissions into such and such databases. And all of a sudden, and this now happens very rapidly, a table is all of a sudden dropped.

16:50Fortunately for us, it's a very similar methodology in terms of detecting that and protecting against that and allowing to recover. but the velocity of that happening is extreme yeah something that i i noted is that like six months ago no one would even discuss with me but a few months ago pretty much every person i meet every uh either in a company tells me either they are afraid of that happening to them or it personally happened to that person who was speaking with me, which is crazy. You see it all over the place. You see real fear from I no longer decide what's really running on my data. I don't know longer understand.

17:40I need to be prepared for both external threats because, you know, all the new models make it much easier for attackers to come to me and attack me. But also from the inside with agents, I actually approved running in my environment. So it's a very, very tricky time. We need to assume breach, whether it's malicious or not, and need to be able to handle it and act accordingly. It's a very weird situation today. Yeah. How do you think about the broader enterprise stack and agents? So, you know, the current stack really evolved around people or humans asking very predefined analytical questions. So we have warehouses, we have dashboards, we have the ETL pipelines, we have BI.

18:32And agents may behave differently and more dynamically. They may be able to reason over much larger sets of data. They may have access to SaaS apps and historical data and a variety of other things in that act. And so what do you think changes in terms of how you store access and interact with data in the context of like the agentic world? Or what else do you think needs to change? Do dashboards go away? Like what shifts? I actually think we'll see more dashboards because this will be the only way to kind of let us figure out what they have going on in the world. Because first, coding agents are starting to write most of the call that's running in the world, so that's indirectly.

19:12But also agents activating other agents would activate other agents and trying to keep track of the non-human identity or that it becomes almost an impossible task. So many actors inside the organization when it's so very hard for a human to understand the change of responsibility. And this is a part of what you're seeing in the proliferation of cybersecurity companies. how many cybersecurity companies you see in NHI, in non-human identity right now, an infinite amount. And there's a reason for that. It became a number one, number two problem right now. In addition to that, second thing is endpoint.

19:54You see endpoint security, which looks like it's sold and so many great companies around it. And just a few years ago, when endpoint was a completely different problem with TBRs, And now everything that's happening, you see people are running agents today on the laptops and the agents sometimes connected to other networks and they are connected to Think and Navy on OpenClaw and connected to your WhatsApp, but also to your internal network and also to other applications. and you see it's very hard for the for the vpo of it's for the cios to understand what should they do on the one hand they want to and they are being pushed pushed by the board by the ceo enable ai in my organization now don't block me you can't block me on the other hand it's so scary i mean every person don't even think of technical people think of the non-technical people building something with, you know, let's say Lovable or any other software that you have for themselves, putting company data there.

21:10They're not even aware for things like security or compliance or who's going to use this data. And they're all using all of those new cool things. So maybe their agent that they're building are using other agents and they're not technical to even understand what it means. So it creates a complete set of actors inside an organization, not bound by the rules of the organization, and not necessarily running within the premises of the organization, but handling sensitive data, which is the property of the organization. It could be exposed to the world. It could be incorrectly used and becomes a big problem.

21:51it's a good thing and bad thing that everyone inside the organization can become builders, whether you're a social media manager, whether you're a film person, whether you're in legal or finance. So it's amazing, but it's also we live in very interesting times in that perspective. How much of the existing data infrastructure do you think survives all this? So, you know, there's all the ETL, data engineering infrastructure that, you know, people have been building and deploying over the last decade. Did that stick around? Does that shift? Does that change? Like, how quickly does all this upend?

22:30So you see there's a strong compelling event to pretty much change everything because the plumbing today is very limited. And everyone built a solution to their set of problems. So think about what happens. Now, Gonen goes downstairs after recording this podcast, and he really wants coffee. So, he goes to the store and buys coffee, and he puts on his credit card. Now, there's a transaction, and this is written in some database somewhere. Okay, right. So, someone needs to, today, what they're doing, they're extracting the data, putting it somewhere, and that's it. Someone else, at some point, takes this data and processes it in some other way, and that's it.

23:12So, there's no connection with all of those stuff, and every person is very different. They don't have the context of what happened before. And the reason it wasn't, and the reason it's very simple. It wasn't so important before to have all the context, all the data for an organization because you could only do with the data things you really intended to do to begin with. So you had a single purpose in your mind when acting on the data. Today, it's very different. Today, you understand that you can collect, if you are able to smartly collect and clean all your data and make sure you start in an efficient manner.

23:48And if you can activate that efficiently, you can let a team go wild with all the data that they have. And the more data that they have and the more high-quality data that they have and the more context, all that data that they have, the team hunting that can create wonders. And think of things which were unimaginable. Let's say that there's one person in an organization who have a list of all the people in New York who love burgers and another person in the organization who has a database of all the people in New York who love pizza. They don't know they can find a list of all the people in New York who love burgers and pizza because they didn't work together.

24:28Now, if you use it for post-training, use it for the new capabilities, you can actually do wonders with that. You can actually start asking intelligent questions. You get the intelligent questions. You can start using it for your own purposes and just something that you couldn't do before. So you're seeing companies, first, they're collecting tons more data than before. The amount of data being ingested is absolutely insane, especially comparing to earlier. We see trends continuously, both us and other companies that we're seeing in data. You see data is growing out of proportions, so much of it.

25:06and so much of it being generated by those new agents. So there's a lot of noise in the data. There's a lot of value and noise as well. So you need tools that are able to both understand data from multiple locations, clean the noise, and make sure all of this data that's been created is actually usable. It doesn't apply with the old tools that were very, some of them were incredible. 5Tron was an incredible company, DBT and so on and so forth, but very niche, very specific tools for that purpose. So this creates a very interesting, brave new world. You've seen companies like Databricks, one of the most incredible companies on the planet, in my opinion.

25:53Look at that. I have more and more data coming in. I don't necessarily know where it is. I'll help you cut all of the data and make use of that. but it's an after effect you already have the data now you need to process that but they are reinventing themselves all the time because they understand that more and more data is being generated by agents and they thought the way I see it is if you can't build them, join them, we'll build our own agents we'll build our own databases they want to take charge of how data is being used, how data is being created and it's completely different than how any other people used that just three or five years ago.

26:39So the goal is really to enable this culture of builders and the culture of agents with the ability to automatically help them understand what's there, automatically help them ingest the data without having to build manual pipelines for each and every application that is being built, and then also help them maintain control on top of the data that's created. Makes sense. And you see with every data that you have, there's another problem right now that lots of data is amazing, but it's scattered, which is a sound of problem. But then you need to access that. You need to pay for that, for storage and, of course, tokens.

27:18And we're not in the time of token maxing anymore. trying to go to actually getting value for every token that we have because it becomes more and more and more expensive. So you want to be very wise in, you don't want, I don't want to say not paying millions, pay millions and even more than that if you need to, but get the value that you can from actually doing so. So it's very expensive, very lucrative. Let's make it relatively as least expensive as you can have it. So I guess, you know, the other thing that you guys have really lived through is the cloud transition. So prior to Eon, you started a company called CloudEndure that was acquired by AWS.

28:06And at AWS, you really saw that migration from on-prem to the cloud at like a huge scale in terms of that big sort of generational shift that had happened before this. how would you compare this infrastructure change to what's happening with AI right now? Like what do you view as sort of the cloud era versus AI era? And what are takeaways or lessons that you can apply across them? Yeah, I think, again, it's like that, but on steroids. And even before we sold our last company, Cloud Underwater AWS, we supported similar large-scale enterprise migrations with other hyperscalers, with Azure and with GCP, where our product was integrated, OEMed into the console.