AI Summit: Ariel Assaraf⁠, Co-Founder and CEO of ⁠Coralogix

Hi everybody, good afternoon, good evening and good morning wherever you are. Thank you very much for joining the Bauden Club today to our AI Summit, featuring executive from Top Tech Company's revolutionizing the world as we speak. My name is Josie Galmazi and I'm joining also by a good friend of mine, Amado Vargas. Amado. - Hard one. Thank you for joining us today. - Yes, so today we are fortunate enough to interview noiner than Ariel Asraf, the co-founder and CEO of Corologix, a company that revolutionizing our organization's monitor, analyze and secure their data by transforming logs, metrics and traces into real-time operational insights. Most recently they raised $115 million of CREC to fuel their AI power to observability capabilities. So Ariel, thank you so much for joining us today. - Thank you very much for having me, Josie and Amado. - Absolutely, they're pleasure is ours. So Ariel, I want to start with a very interesting question about the competitive landscape and the early days because you started building the company knowing from the outset that you would compete with giants like Datado, New Relic and Splank. What was your attitude back then and how did it affect building the company in the early days? - I think we learn a lot of lessons doing that. I can say we thoroughly thought about it before we started. I think our assumption was we will just build something nice and fun to use and people will use that. We didn't really understand enough about the motion of enterprise sales and it's something that took us actually a few years to get to. So before we started, we did what all the accelerators and experts tell you to do and we started interviewing a bunch of customers. So we went and talked to maybe 50 or 60 customers and we just asked the wrong questions. Now maybe we were fortunate to do that because of that we actually started the company and turned out well. At the beginning our questions were mainly around would you like to use what we're building? And so you come to a customer and you'd ask a question like, hey, are you looking for a system that's more easy to use and friendly? Sure. Do you want something that will automatically analyze data in surface anomalies for sure? Do you feel like you're paying too much for your current tools? Yes. So, you know, everything was kind of green light and we said, okay, there's big opportunity to market. But as we went to the market, we learned that we were missing a few key questions like, what is it about the current tool you're using that you cannot give up? How many people are inside the organization and what does it take to replace? We'll be willing to pay for more than one tool if the other one give you additional value. What is the cost of change for you? And not asking those questions meant that when we started going to market early 2016, I mean, we got a lot of POCs, but very, very few customers actually converted because most of them were just like not understanding, how will we replace that massive product suite that they currently have? What will be the cost of change? How will we support them through that? How will they trust a small company with their data? And the net is we got to early 2019 or late 2018 with almost no revenue or customers. Well, at that point in time, we learned that we're going to have to come up with something completely different than our competition that meets the customers where they are from a pain point of view. It doesn't try to kind of say, hey, here's something better and hope for them to come to where you are. But actually, we've taken a lot of that experience that we've gathered throughout the years and we understood that cost is a big factor. The level of support and partnership are a big factor. The real timeness of insights and the scale are real issues. And we decided to redesign correlgics around the technology that we call stream, a back then the main core of correlgics, which was how do we analyze insights in real time or data and real time without having to use storage? So it sounds a bit controversial in terms of people need that long term stuff. So how do you not store it? And then you give them insights only based on real time data. But the whole point around analytics is how do you provide long term insights and correlation? And so we understood that we're going to have to create something that doesn't exist today, which is what we call stateful streaming. How do we keep the long term knowledge we have on data while not storing it? And that's how we created streaming. When we went with that to the market, by the way, needless to say at that point in time, the company's almost shutting down because I mean, we're, I'd say mid 2018, almost no revenue or customers, not a lot of money in the bank. We're seven people at the time. So it's kind of our last bullet. And we managed to around the third or fourth quarter get to around $50,000 of MRR, I think, within just a few quick months. And the board sees that and we ran out of money and basically the board extends our run rate by a couple million bucks back then. And when we hit the end of 2018 into early 2019, we're already crossing the one 1.2 million dollars of ARR. At that point in time, when we understood we have something in our hand. So, you know, to your question, we didn't really understand the competitive landscape. It took us a few years. And then when we actually hit that wall with numerous customers not being able to move to correlatics, we understood that we have to come up with something completely different from the technology standpoint and not just kind of repackaged the existing capabilities in the market. It seems it's all about the mindset of people team when necessary in order to really understand the customers and good market fit. And it's very interesting what you said because fast forward to 2019, you already got a position of 40 million dollars, which is not bad. And as you mentioned, you only started to get back then the traction that what we see today is corologic. But what made you sure that this is only the beginning of corologics and you're going to be catapulted to your not a level? It's interesting actually. It's a good lesson for acquires that we've also learned that once we've acquired a company, we just started getting traction. So, you know, in our view, obviously, it was a great opportunity this company just barely shut down six months ago. Suddenly, I was not even 30 at the time, we could be millionaires. So we obviously said yes to the offer and we signed the term sheet. Then the amount of scrutiny and due diligence that much of it was unnecessary was just huge. And for months and months, it kept dragging from 45 days to 60 to 90 days closing time. But what happened is that during the time, corologic continues to grow tremendously well. So we're at like 1.2 million at the point in time where the SBA was rated as signatures, we were already at 2 million. And so I was looking at my co-founder and saying, hey, look, I mean, I think we can end this year at around $3.5 million. Why would we sell the company at that growth rate with this opportunity and this market for this amount? And days before signatures, we just went back and said, guys, we're not going to do it. The problem was we didn't have money in the bank. So because it extended and extended, we actually didn't have money to run the company at that point in time. We went to the investors that we knew I live back then, we're close to corologics. And we said, we got a couple days before our board that needs to approve the SBA for signatures. But we don't want to sell. Can you give us a term sheet? And he said, listen, okay, I'm happy we'd like the company. I want to run you through my partners and then I need you to come to our office. I said, no, you're not getting it. It was Aaron Rosenstein back then. I don't think you get it. I got a couple days I need a term sheet. So he wrote that term sheet on WhatsApp and sent me a text and said, okay, you can show that to your board. I just need to approve it in our committee. So we're back, we're in the US, we're back next week. I said, I can't wait. So we did it while they were in this room in the airport barely hearing us, but they approved it and then I went to the board and said, hey, I don't want to sell. I want to take this offer. And he sent the text actually saying, tell them we're never backing off of it. So don't worry about skipping the SPA. Style from stage one immediately, he was on the committee of the M&A. He immediately resigned from the committee and said, okay, I'm with the team. We don't want to sell. And then we didn't. And I live came in with $10 million at the time. That was our A round. I remember Aaron telling me, this is the hardest round you'll have. At this point, I was just continuing executing and things will get easier. So up until I live, we raised like three and a half million dollars. I live was 10 and since then we raised another, I want to say, $350 million. So he was probably right about that being the most difficult one. Absolutely. It seems like a fairy tale indeed. And I want now to go a little bit into the technical aspect of building co-logics because you've written about building observability, not as a devops, checkbooks, but rather as a culture, with ownership across teams, right? So in your experience, what was the other part of organizations doing, demilize observability inside the organization as a muscle memory, rather than as an after foot? It starts with a question is observability just a fancy name for monitoring or is it really something else? Monitoring means you build something, you put it in production, you monitor it to make sure it works. Observability is a cultural mindset that says, I need observability into what I have running in production, doesn't even matter if it works or not, even if everything works. I need observability to understand what's happening. I need observability to understand how my software is performing, how my customers are engaging, how my infrastructure is efficient. how my security is up to standard and how my compliance meets their needs. That mindset is how do you create shared ownership of all these big aspects that I just discussed right? The quality, the stability, the customer experience, the security, the compliance, how do we create that cross responsibility? And this is what we later, by the way, design as the observability maturity model. The correlagious ones with the customer is the footprint that we've seen across newer customers when we define four stages of observability culture within the organization. The first one is telemetry. How do you get through actually collecting the proper telemetry that you need? Throughout the organization, how do you make sure that you have all that data, that audit data, the compliance data, the user information all the way from front end, the backend information, the metrics, but not just infrastructure, but also the custom metrics that represent the business logic of your organization. And then how do you go to what we call engineering level intelligence, meaning how do we let me know? That's the basic step. How do we let engineers monitor and fix issues faster? But from there, there are two big steps that are more organizational. One is what we call production level intelligence. How do we map out the critical workflows that the company is responsible for? The critical workflows that our customers need to have working 100% at a time, not even 99%. And how do we understand where the people responsible for them? Meaning engineers, the different services that are deploying them, the infrastructure that's running them, the platform teams that run their CI/CD, the product managers that make changes, the support engineers that support them, the QA engineers that release the production, the security folks that make sure that they're secure. And how do we map out all those critical workflows in the organization and have an agreement with them? There's share responsibility. And there are singular dashboards and alerts and views that everyone reads from. So we create a common language around these critical transaction. So when people look at when the BI team looks at one dashboard and the ability team looks at another and then the product team looks at a product analytics tools out there, it's really hard. It's like the Babylon power. How do you create one language? And someone says, I think something's wrong. Everyone all hands on deck are handling it. So that production level intelligence is mapping those out, creating the shared responsibility model and then routing those alerts and reports to the right people about any incident. So people that need to act know that they need to act and then people that need to be aware, aware. And the last phase, the fourth phase of these are really maturity models, what we call business level intelligence. How do we take that data? And the end vision for Corel Laws is what we say help businesses make business decisions. It's not about monitoring or uptime at all. It's about this information embeds everything that happens within the business. So if you're an online business and there are two types of business in the world, there's physical businesses that are enabled by technology. You could put automotive, energy, retail, all these inside. There's technology supporting the business, but the business is physical in essence. There are business like us and e-conners and FinTech that are completely online. Everything that happens within my business is recorded and the observability data and digital data. So if you think about a BI event, about a user tapping into his app, you'll say a user, a UOC tapped into the app and wired funds, right? Think about the observability data that I have around it. I have the CDN event, the WAF event, and the security score for that user. I have his click. I have a recorded session of him. I have how much time it took did it succeed or not through what databases it went, what queries did it run, everything, the cost of info for that particular action, everything is recorded. So I have so many dimensions and such better understanding of my business from observability data than BI gives me that we strive to go through those three phases into business level intelligence where the management is actually looking at these dashboards. And we got a few really interesting case studies, one that we've done with Puma, for instance, the apparel company and how they drive marketing decisions, campaign decisions, inventory decisions through the core logics application, how users interact, what experience they're getting, is the website ready for a campaign or not. Are we at a software version that will enable a surge of users if I do something right now and will serve them the right way so I capture the entire value from that campaign. So once we get to that fourth stage, we actually power up the entire business and this is where core logics is trying to get. - That's super interesting. And you mentioned also that there are the physical businesses and the internet based businesses, but I was wondering, let's say you started to grow all over the world, different continents, different kind of businesses. I guess something that in America, it's not the same as in APAC or in the app. So from regulatory standpoint, or maybe even compliance challenges, have you faced any special challenges while growing so much on the world? - Data is an infrastructure. It's not a product. And when you sell infrastructure, compliance and operations are a big, big part of your company. Core Logics platform teams are huge because they operate 10 different regions in APAC, like you said, and Imiya, and both East and West US. The idea is how do we keep data local? How do we adhere to all the different compliance? By the way, everything from SOC to type two of user information, the PCI of financial information. So core logics, it's not that we recommend to do that, but by law, core logics is even capable of storing credit card numbers because we comply with PCI. Hey, but compliance, we are technically able to store medical records for users, even though it's not the purpose of the system, but you'd never know what the customer is going to send you. So we have to adhere with all these, even by the way, the new compliance for AI, ISO 42, 0-0-1, or logics, it's compliant with that. The federal compliance fed ramp is a huge lift that we went through so we can serve federal clients. That is the first technology step to build that infrastructure. Now there's a human factor to it because such a critical component, the car is so much trust you got to have local teams. So we have an office in India, we have an office in Singapore, we have an office in London, we have an office in Boston and in San Francisco. Obviously a big office until you get the idea was, how do we get that global spread? And when we grew the company, we thought about it this way, we said, okay, there's two ways. Either you just focus in the US and then you try to spearhead other territories, which a lot of companies do, they start only US, and then as they think about expanding or going global or maybe going public, they start trying to open up other regions. For us, we went very thin, all regions, and then now we're kind of beefing them up with more teams and more customers and more presence. So, a power logic is operating and selling today, significant numbers, middle east, Europe, UK, if you separate Europe, UK, APEC, Southeast Asia, and the US, and even South America. So, I think that decision will pay off because now it's going to be a lot easier for us to get to that next phase of, you know, $500 million dollar revenue, billion dollar revenue and become a public company. And I guess now with the rise of AI, it could cut up to you to another level with all the capabilities that you actually just recently launched, right? Because just a couple of months ago, Corelogic launched AI Center and AI agents named Oli to help answer production, the questions among other features of course, but can you share with us a little bit about this platform and what architectural or product challenges did you face in building an AI agent directly on a observability data? - It's a great question. I think a lot of the work that we've done today was to enable that agent. The fact that we store the data, we format it in stream, and then we store it in the customer's own storage for remote queries. That's the core of Corelogic. It's a streamer, the streaming engine and data prime, the remote query engine. So the data sits on the customer's, there's this three, in readable format. That format is basically AI ready. When the agent runs on that data, it uses our formatting engine in stream, it uses our query language, which is declarative and very powerful. It uses our infinite retention because data is stored on the customer's archive. It uses the amount and variety of data you can set to Corelogic's both commercially and technically we enable that. And it also uses the cardinality power of Corelogic. It's capable of querying very high variance of data. This is one of the biggest challenges for query engines. So if you got a million unique values inside one column, we're able to extract from this. So the AI agent is actually running on an infrastructure that was completely built for that manner. And then when you think about the previous question, the questions you'd ask all of you are not, hey, this user has failed, how can you help me? It works for that. But the big questions would be, which one of my customers is having a bad day today? Why is that and how can I resolve it? So those are questions that you'd ask an actual SR engineer. And we did a lot. We, all these part of the acquisition that we've done of a poor year earlier this year, we bought that team, we built out the AI Center, which is an excellent center of engineers. And they are responsible for deploying all the width, the small models that supported the personas that were building for it. And obviously all the fine tuning and evals to make sure that it's accurate and how it's accurate. answering questions. So that's a big one for us that I think is going to take over a lot of the new business that we're going to land, I'd say from mid-26. Actually, mention Afoya, so it's a great segue for my next and last question before I pass the mic to Amato. So you acquired a company back in December 2024 to build your AI observability and gallery capabilities, but from your perspective, when you acquired a company like Afoya, I'll be approached the dilemma of integrating the products right away to the platform versus continue selling it separately. Our approach was we're in a market that you talked about a lot of big players. It's all a platform play. It's all in one versus best agreed, best agreed, lost in our space. People expect one platform to deal with everything. So when we bought a Portia, the first thing we've done actually was we sat down, we took difficult decisions on what products were discontinuing with Apoya, what products are combining with the Chorologics vision. We took a very tight timeline to embed them within Chorologics three months. We made it happen, and then we took a big chunk of that team and made it work on the vision. So the idea was, Apoya as a brand, as a product, as a product line will not exist within a quarter. It is all part of the re-platform that we'll do to serve the Chorologics line as one of the multi-products that are connected together. Now, every company you hear about, they create a product. Some companies have an opportunity to create multi-products because they got multi-sellers or multi-issues that they're solving. We're there now. We have multiple products. Very few companies and observables is one of that spaces. Allow you to build a platform where like Chor services feed the entire organ. Now, AI, the agenda clear, is going to be that glue where you ask a question without thinking about what product you're asking. It can be the logs, the metrics, the APM, the realism monitoring, the AI observability. You just ask the question like I said, what is causing my customer's frustration and how to solve it? I didn't say check the spans, check the metrics, check the logs, check security data, it doesn't matter. The idea is to have just a platform and an overlay of intelligence that allows you to interact with the data even if you're not an expert in a way that drives business outcome and not just looks at the downside. Is software bad and we're just preventing it from being broken or is software great and we're going to extract as much information and value from it to drive our business. So we obviously took that second approach. The super interesting and actually a bunch of follow-up about it, but certainly we have to move on. So, I thank you so much and Maddo, without the further ado, the floor is the other side. How do you as a founder, you brought up three points that I found interesting. Maintaining back growth rate, you also had a time where you wanted to lay out a term sheet on what's that. But also, I think the most interesting thing was you weren't looking to repackage current capabilities already on the market and I think that's why CoreLegix has been so successful. More on that in terms of your mentality as a leader, you talked about linking observability to business logic, customer experience and revenue metrics. Do you see a future where observability is as native to leadership dashboards as it is to engineering? It has to happen in our view because we look at the world we're getting into those type of physical businesses. They're going to get more and more and more online. If leadership at the most traditional companies don't become tech savvy within five years, they're just not going to fit their business, they're just not going to serve their business. These guys have to understand that when a bank, when a car manufacturer, when an energy company is selling, the digital experience they provide to their customers, the resiliency, the security is a very, very big part of their decision. So, you can run campaigns, you can make acquisitions, you can get into new markets. If you're not excelling the technology piece, your business is not going to work. I think one of the best examples is, I mean, you just look at Tesla, for instance, as a company disrupting a very traditional industry, it's all about the software. It's all about the Tesla software. It's not really about, you got so many great electric car manufacturers in China today. It's all about the app and the software running the car and how quickly it evolves and how reliant it is. I'm not even talking about autonomous car. Just a simple user experience and resiliency and customer, the lightness that these cars bring and it's going to happen throughout all businesses. So, I'd say everyone are talking about, yeah, data is the biggest asset of the organization and data is the new oil. Everyone has these things. But then they look at this tiny static data set of BI with 20 different columns that they pre-define with specific labeling that gives them five graphs. While observability is an infinite treasure of unstructured data that, like I said, records every single user interaction from 20 different dimensions. And I think it's on us to make it available. I mean, Ollie is one of the answers, but there are plenty more. And it's on them to learn how they interact with it so they can actually make better business decisions for their companies. I'm a certain and when you spoke on the unstructured data, I think that really connects with what you told you about how you view data as an infrastructure and not as a product. And so that makes a lot of sense. Looking into AI governance, as AI becomes core in many systems, observability will likely play a role in ensuring reliability, fairness, and compliance. How do you plan to end guardrails, auditability, and transparency and observability for AI systems? It's a great question. I think there's balance here to fine. So we offer guardrails, by the way. I won't get too much into technical details, but the idea is there's a small language model that acts as a judge that model is only trained for that guardrail. So if your guardrail is, I don't know, don't be racist. Any message coming in and out of the model will go through the model who's an racism expert. I don't know if it sounds great. It'll look at the question and say, hey, this is okay. And the answer, hey, this is a care not or guardrail. But those are technical details. I think the interesting part is what is the balance between guardrails, security, and compliance, and privacy, and completely destroying the value of AI? So I was on a call with one of the larger banks in the US. And I asked, how are you guys planning to protect a lot of your agents or using AI now? And they say, no, no, no, they have a very strict list of questions that they can ask, and they know they can't ask anything else. I said, why do you need AI? Just give them a handbook with these questions and answer. And it works. AI is just less predictable. It's a great example on how you actually killed a value of something because you added so many guardrails and so much process to it. But on the other hand, how do you understand what is right and wrong and what is hallucination and what is true? So I'd say the big change here is if software was working and not working, right? There's like green and red. AI is this gradient of options between completely problematic to an amazing answer. And every organization will need to put the dot where they want to go. I would put the dot like right here, right before terrible. But some organizations will go here and say only really good answers. The idea is to understand that AI is not software. You cannot monitor AI like your monitoring software. You cannot measure or judge AI like your judging software. If you asked a model to, I don't know, give you assistance on your bank account and it just told you to it thinks that you should buy Nvidia stock. Is this bad or good? Maybe it's a great advice. And the week later, the stock's up 50%. Maybe it's terrible advice since down 30. Maybe it shouldn't even recommend that. But do I tell the AI models you never recommend anything? Because there are things that I want them to recommend. It's a lot more complex in software and software. I'll say it should recommend things or not. And it's very straightforward. There's a code that needs to do something and when it recommends needs to have this output. And now we're stepping into a non-deterministic world that we need to embrace and we need to train our people. By the way, I'm saying our people like I'm thinking about employees, I would start training from third grade. How do you guys interact with AI in a way that is safe in a way that adds guardrails to it by your behavior and your judgment and not the software? And it's about how do you learn how to extract the maximum value from this really powerful tool you have rather than limiting the tool and what value it can provide to you because every person and every use case is so different that if you try to protect all the different people in use cases from AI, again, you end up with a list of questions and answers. So that would be my perspective of this. And again, we did the whole 42 zeros. You're one of compliance and obviously data leakage and hallucinations as much as you can prevent them. All these things are critical. But at the end, the value will come from educating and training people and how to adjust to this new world. Fast in the answer that that's the problem I've heard. So in separate AI from software and so I think it's a really interesting way to look at it and how software is more straightforward and AI can't be configured and judged. So as first of all, I heard that, but I think that for our viewers looking into the future, those want to master AI, don't look at it from a software mentality. It's not just copy and paste. It's completely different intellectual domain. So that's a great answer. Looking into it, advice for founders, because you were a founder who had relatively early success. If there's an observability founder out there, building their company right now, what's the one thing they should think of from day one? Product market. I can't say don't start. I have to say another. Okay. If I was stepping into observability, I would try to identify what is a trend that will be significant in the world three, four years from now and start building towards that. It takes quite a few years to build an observability company so it can actually sell. And so I would try not to build for today. I would try to build for the future. And if you look at the successful data companies, Splunk started 2004, building for the big data revolution that only came in 2010. And obviously, Splunk did really well. The analog started 2010, 11, building for the cloud way before people really on the cloud and actually did really well. I wanted to say, where all logic started 2015 and built for the world of AI. Like I said, all the infrastructure and how we started data and hopefully, you know, that opportunity we see as bigger than the cloud and big data combined. And then if you're starting today, what is what is that that you want to build for? That would be my advice. Yeah, great advice. I believe you've been like Nvidia began in the late 90s. And so they they were 20, 25 years ahead of the game before that big explosion. Yeah. And they were just ready to harvest all that value, right? Because they were there. So definitely if someone started the chip company today to build for AI, maybe they do it. I mean, maybe they make it well, but they're never going to be in the video 1000 percent. Heading more into audio on the mentorship side of things. Was there a particular mentor mentors that impacted you most and all the key lesson that stuck with you on the journey? I don't want to forget anyone. But I'll say three people that I think are have been transformative to my career. Obviously, all the investors and everyone believe in correlating the totally some of the stories, giving up the acquisition and taking that term sheet and all these things. But that aside, three people that really helped me out throughout the years. One is Moshe Liechtmann. He's an Israeli investor, former Microsoft person. He was the project manager. I hope I'm not wrong. He was a project manager for Windows 95, you know, think of that role. Basically, he led Microsoft 95. And then when he moved to Israel, they open up the Microsoft Center in Israel around him. It's now tens of billions in revenue. It's like one of the biggest ones in the world. And he was an early investor in chronologics. When things went really wrong in the beginning, he told us, "Guys, you're the third generation of Israel. There was the first of seconds. It's funny that he recognized back them like the Splunk, the Data Dog. You guys are the third generation. And I think you're going to succeed. Keep going." Second one is Mike Gregor on our board. He's joined us. He was the CEO of CA, of Taleo, of PeopleSoft. He's joined us in 2022. And now he's chairman of the board. And I meet with him every couple weeks. He's been hugely helpful and just kind of navigating through those phases of scale. And what does it mean to sell to enterprises? What does it mean to manage a company? How do you grow as a leader? And the third one also joined us in 2022. It's Mark McLaughlin. It used to be a CEO of Palo Alto. Now the chairman of Qualcomm also an operating partner at Advan, our investor in 22, that I meet every couple of weeks. And you know, every advice his guy gives is just cool. He's ability to hear out a problem and then analyze it on the fly and give you an answer. As if he's been in your world for a year now. It's something that I haven't seen before. So I'd say plenty of people, but those three are definitely at the top. And just to add something, I also, Mark is going to join us in one week as well to one of the decision. Mark or Mike? Mark, do the one who Palo Alto now determine of Qualcomm. Okay. So you guys, I'm ashamed that my, my episode is going to be one before him, but okay. Awesome. So when it comes to, there's a lot of friction in terms of being a founder, you know, the high intensity environment. What keeps you motivated during a long cycle is in the inevitable setbacks? And that's a really good question. And if you don't ask yourself this question, then you're an auto mode. You're not really thinking, but why am I doing this is a big question, right? Like, why am I doing this? Why am I doing this now? And I think what started as I want to build something and continue as I want to create some success and then continue as I want to make some money and then continue as I want to have the experience of building a big company, I think at this point, you get above a certain level of size and, you know, vision. It's more around what is it that we want to leave in this world as a company? Like, how do we create a generational company that will be here 20, 30, 50 years from now? How do we create a company that will be named as something that's changed something into technology landscape, especially in the big market, like observability? So that, that would be the thing that drives me now and I'm whenever we set dreams, like I said about our vision, it's, it's transform how businesses make decisions, you know, imagine in 20 years, people say, Hey, a quarrelogics is the company who's transformed how businesses make decisions and turn it into something that's based on machine data and not business data. So obviously, that's a big dream for us. I'm not sure we're ever going to make it, but it's a good enough reason to keep going. Well, you guys are on the right track. So I can see it happening. I hope. One last thing on observability and potential alternate career path, because obviously, you've mastered the domain. However, if you weren't in the observability space, is there enough profession that you'd have in mind in an alternate universe? I'd probably be a doctor. That's that, that would be my dream. Like, I always thought I'm going to, I'm going to study medicine and, you know, somehow with the 80 to 100 route and then math and university, I ended up, it's starting a company, but definitely that would probably be been my parallel universe. And then the final question before paying back to OC in terms of your own personal time, is there a special hobby or favorite movie or television series that helps you in line? Actually not. No, I just like spending time with my family. I got three little girls whenever I have some free time. That's that's what I like to do. And yeah, no, no hobbies here. I used to play basketball, but it's been a while since then. I always said no, family's the best unwind. So, we know, it's hard at work and then relax at home, that's the way to do it. Arielle, thank you very much for the great answers and back to you'll see. Thank you very much, Arielle. Yeah, so Arielle, thank you so much. Actually, I think the NBA starts today. So, it's definitely interesting time for basketball fans out there. So, Arielle, thank you so much again for joining us. It's really been a pleasure, oh, seeing you and learning from your vast experience. And for everybody else in the audience of Stay Tuned, we've added amazing episodes from one of the mentors, Rafael, Mark, Chairman of Qualcomm. We have the CEO of Tabula, CEO of Box today. We are going to release it and yeah, it's going to be interesting and I thank you so much. Thank you very much for having me. I really enjoyed it.